Автоматически выберите модель классификации оптимизированными гиперпараметрами
Учитывая предиктор и данные об ответе, fitcauto автоматически пробует выбор типов модели классификации с различными гиперзначениями параметров. По умолчанию функция использует Байесовую оптимизацию, чтобы выбрать модели и их гиперзначения параметров, и вычисляет ошибку классификации перекрестных проверок для каждой модели. После того, как оптимизация завершена, fitcauto возвращает модель, обученную на целом наборе данных, который, как ожидают, лучше всего классифицирует новые данные. Можно использовать predict и loss объектные функции возвращенной модели, чтобы классифицировать новые данные и вычислить ошибку классификации наборов тестов, соответственно.
Использование fitcauto когда вы не уверены, который вводит классификатор, лучше всего удовлетворяют вашим данным. Для получения информации об альтернативных методах для настройки гиперпараметров моделей классификации см. альтернативную функциональность.
Если ваши данные содержат более чем 10 000 наблюдений, рассмотрите использование асинхронного последовательного алгоритма сокращения вдвое (ASHA) вместо Байесовой оптимизации, когда вы запускаетесь fitcauto. Оптимизация ASHA часто находит хорошие решения быстрее, чем Байесова оптимизация для наборов данных со многими наблюдениями.
Mdl = fitcauto(Tbl,ResponseVarName)Mdl классификации настроенными гиперпараметрами. Таблица Tbl содержит переменные предикторы и переменную отклика, где ResponseVarName имя переменной отклика.
Mdl = fitcauto(___,Name,Value)HyperparameterOptimizationOptions аргумент значения имени, чтобы задать, использовать ли Байесовую оптимизацию (значение по умолчанию) или асинхронный последовательный алгоритм сокращения вдвое (ASHA). Чтобы использовать оптимизацию ASHA, задайте "HyperparameterOptimizationOptions",struct("Optimizer","asha"). Можно включать дополнительные поля в структуру, чтобы управлять другими аспектами оптимизации.
[ также возвращает Mdl,OptimizationResults] = fitcauto(___)OptimizationResults, который содержит результаты выбора модели и настраивающего процесса гиперпараметра. Этим выходом является BayesianOptimization возразите, когда вы используете Байесовую оптимизацию и таблицу, когда вы используете оптимизацию ASHA.
Используйте fitcauto автоматически выбрать модель классификации оптимизированными гиперпараметрами, учитывая предиктор и данные об ответе, хранимые в таблице.
Загрузка данных
Загрузите carbig набор данных, который содержит измерения автомобилей, сделанных в 1970-х и в начале 1980-х.
load carbigКатегоризируйте автомобили на основе того, были ли они сделаны в США.
Origin = categorical(cellstr(Origin)); Origin = mergecats(Origin,["France","Japan","Germany", ... "Sweden","Italy","England"],"NotUSA");
Составьте таблицу, содержащую переменные предикторы Acceleration, Displacement, и так далее, а также переменная отклика Origin.
cars = table(Acceleration,Displacement,Horsepower, ...
Model_Year,MPG,Weight,Origin);Данные о разделе
Разделите данные в наборы обучающих данных и наборы тестов. Используйте приблизительно 80% наблюдений для выбора модели и настраивающего процесса гиперпараметра, и 20% наблюдений, чтобы проверить производительность итоговой модели, возвращенной fitcauto. Используйте cvpartition разделить данные.
rng("default") % For reproducibility of the data partition c = cvpartition(Origin,"Holdout",0.2); trainingIdx = training(c); % Training set indices carsTrain = cars(trainingIdx,:); testIdx = test(c); % Test set indices carsTest = cars(testIdx,:);
Запустите fitcauto
Передайте обучающие данные fitcauto. По умолчанию, fitcauto определяет соответствующие типы модели, чтобы попробовать, использует Байесовую оптимизацию, чтобы найти хорошие гиперзначения параметров и возвращает обученную модель Mdl с лучшей ожидаемой эффективностью. Кроме того, fitcauto предоставляет график оптимизации и итеративное отображение результатов оптимизации. Для получения дополнительной информации о том, как интерпретировать эти результаты, смотрите Многословное Отображение.
Ожидайте, что этот процесс займет время. Чтобы ускорить процесс оптимизации, полагайте, что определение запускает оптимизацию параллельно, если у вас есть лицензия Parallel Computing Toolbox™. Для этого передайте "HyperparameterOptimizationOptions",struct("UseParallel",true) к fitcauto в качестве аргумента значения имени.
Mdl = fitcauto(carsTrain,"Origin");Warning: It is recommended that you first standardize all numeric predictors when optimizing the Naive Bayes 'Width' parameter. Ignore this warning if you have done that.
Learner types to explore: ensemble, knn, nb, svm, tree Total iterations (MaxObjectiveEvaluations): 150 Total time (MaxTime): Inf |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 1 | Best | 0.14154 | 15.261 | 0.14154 | 0.14154 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 201 | | | | | | | | | MinLeafSize: 7 | | 2 | Accept | 0.18269 | 0.78311 | 0.14154 | 0.14154 | knn | NumNeighbors: 3 | | 3 | Accept | 0.23397 | 0.12849 | 0.14154 | 0.14154 | knn | NumNeighbors: 91 | | 4 | Best | 0.092308 | 8.1968 | 0.092308 | 0.11151 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 274 | | | | | | | | | MinLeafSize: 15 | | 5 | Accept | 0.20833 | 0.122 | 0.092308 | 0.11151 | knn | NumNeighbors: 4 | | 6 | Accept | 0.22115 | 0.074924 | 0.092308 | 0.11151 | knn | NumNeighbors: 28 | | 7 | Accept | 0.16923 | 0.22742 | 0.092308 | 0.11151 | tree | MinLeafSize: 105 | | 8 | Accept | 0.37179 | 0.59424 | 0.092308 | 0.11151 | svm | BoxConstraint: 0.022186 | | | | | | | | | KernelScale: 0.085527 | | 9 | Accept | 0.37179 | 0.095134 | 0.092308 | 0.11151 | svm | BoxConstraint: 0.045899 | | | | | | | | | KernelScale: 0.0024758 | | 10 | Accept | 0.24615 | 0.92999 | 0.092308 | 0.11151 | nb | DistributionNames: kernel | | | | | | | | | Width: 1.1327 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 11 | Accept | 0.16923 | 0.072747 | 0.092308 | 0.11151 | tree | MinLeafSize: 78 | | 12 | Accept | 0.26923 | 0.098838 | 0.092308 | 0.11151 | svm | BoxConstraint: 11.063 | | | | | | | | | KernelScale: 15.114 | | 13 | Accept | 0.12923 | 0.09922 | 0.092308 | 0.11151 | tree | MinLeafSize: 3 | | 14 | Accept | 0.21154 | 0.077411 | 0.092308 | 0.11151 | knn | NumNeighbors: 2 | | 15 | Accept | 0.14154 | 0.070449 | 0.092308 | 0.11151 | tree | MinLeafSize: 1 | | 16 | Accept | 0.14769 | 0.07596 | 0.092308 | 0.11151 | tree | MinLeafSize: 2 | | 17 | Accept | 0.14154 | 8.0622 | 0.092308 | 0.12513 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 208 | | | | | | | | | MinLeafSize: 10 | | 18 | Accept | 0.37179 | 0.092223 | 0.092308 | 0.12513 | svm | BoxConstraint: 116.46 | | | | | | | | | KernelScale: 0.52908 | | 19 | Accept | 0.22769 | 0.14447 | 0.092308 | 0.12513 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 20 | Accept | 0.22115 | 0.063369 | 0.092308 | 0.12513 | knn | NumNeighbors: 8 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 21 | Accept | 0.37179 | 0.11033 | 0.092308 | 0.12513 | svm | BoxConstraint: 45.341 | | | | | | | | | KernelScale: 0.76949 | | 22 | Accept | 0.12923 | 0.069586 | 0.092308 | 0.12513 | tree | MinLeafSize: 3 | | 23 | Accept | 0.10154 | 0.063197 | 0.092308 | 0.12513 | tree | MinLeafSize: 5 | | 24 | Accept | 0.22769 | 0.18312 | 0.092308 | 0.12513 | nb | DistributionNames: kernel | | | | | | | | | Width: 0.42571 | | 25 | Accept | 0.11385 | 0.062328 | 0.092308 | 0.12513 | tree | MinLeafSize: 11 | | 26 | Accept | 0.13782 | 0.069932 | 0.092308 | 0.12513 | svm | BoxConstraint: 9.7286 | | | | | | | | | KernelScale: 293.41 | | 27 | Accept | 0.22769 | 0.053256 | 0.092308 | 0.12513 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 28 | Accept | 0.21795 | 0.062882 | 0.092308 | 0.12513 | knn | NumNeighbors: 42 | | 29 | Accept | 0.24308 | 0.20538 | 0.092308 | 0.12513 | nb | DistributionNames: kernel | | | | | | | | | Width: 4.4662 | | 30 | Accept | 0.37231 | 6.3057 | 0.092308 | 0.1289 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 267 | | | | | | | | | MinLeafSize: 131 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 31 | Accept | 0.24308 | 0.18689 | 0.092308 | 0.1289 | nb | DistributionNames: kernel | | | | | | | | | Width: 0.66296 | | 32 | Accept | 0.22115 | 0.070866 | 0.092308 | 0.1289 | knn | NumNeighbors: 28 | | 33 | Accept | 0.13846 | 0.080818 | 0.092308 | 0.12465 | tree | MinLeafSize: 25 | | 34 | Accept | 0.21474 | 0.061727 | 0.092308 | 0.12465 | knn | NumNeighbors: 14 | | 35 | Best | 0.089231 | 5.4041 | 0.089231 | 0.12465 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 215 | | | | | | | | | MinLeafSize: 13 | | 36 | Accept | 0.14154 | 9.5 | 0.089231 | 0.12465 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 254 | | | | | | | | | MinLeafSize: 31 | | 37 | Accept | 0.22769 | 0.058919 | 0.089231 | 0.12465 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 38 | Accept | 0.37179 | 0.07729 | 0.089231 | 0.12465 | svm | BoxConstraint: 0.0073633 | | | | | | | | | KernelScale: 774.33 | | 39 | Accept | 0.16923 | 0.066123 | 0.089231 | 0.12552 | tree | MinLeafSize: 82 | | 40 | Accept | 0.20833 | 0.063808 | 0.089231 | 0.12552 | knn | NumNeighbors: 4 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 41 | Accept | 0.37231 | 6.3806 | 0.089231 | 0.12552 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 274 | | | | | | | | | MinLeafSize: 150 | | 42 | Accept | 0.22462 | 0.1782 | 0.089231 | 0.12552 | nb | DistributionNames: kernel | | | | | | | | | Width: 121.64 | | 43 | Accept | 0.20308 | 8.3787 | 0.089231 | 0.12552 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 229 | | | | | | | | | MinLeafSize: 117 | | 44 | Accept | 0.16923 | 0.060332 | 0.089231 | 0.12291 | tree | MinLeafSize: 84 | | 45 | Accept | 0.22769 | 0.054836 | 0.089231 | 0.12291 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 46 | Accept | 0.22769 | 0.051984 | 0.089231 | 0.12291 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 47 | Accept | 0.092308 | 5.064 | 0.089231 | 0.11792 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 212 | | | | | | | | | MinLeafSize: 49 | | 48 | Accept | 0.14769 | 10.667 | 0.089231 | 0.11634 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 288 | | | | | | | | | MinLeafSize: 25 | | 49 | Accept | 0.23077 | 0.17743 | 0.089231 | 0.11634 | nb | DistributionNames: kernel | | | | | | | | | Width: 73.249 | | 50 | Accept | 0.37179 | 0.12425 | 0.089231 | 0.11634 | svm | BoxConstraint: 0.0036501 | | | | | | | | | KernelScale: 1.0504 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 51 | Accept | 0.21474 | 0.083155 | 0.089231 | 0.11634 | svm | BoxConstraint: 64.859 | | | | | | | | | KernelScale: 23.779 | | 52 | Accept | 0.37179 | 0.082077 | 0.089231 | 0.11634 | svm | BoxConstraint: 0.16622 | | | | | | | | | KernelScale: 4.4901 | | 53 | Accept | 0.25846 | 0.21018 | 0.089231 | 0.11634 | nb | DistributionNames: kernel | | | | | | | | | Width: 0.079498 | | 54 | Accept | 0.21154 | 0.063584 | 0.089231 | 0.11634 | knn | NumNeighbors: 2 | | 55 | Accept | 0.13846 | 9.0089 | 0.089231 | 0.10262 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 234 | | | | | | | | | MinLeafSize: 8 | | 56 | Accept | 0.36538 | 0.086614 | 0.089231 | 0.10262 | svm | BoxConstraint: 271.6 | | | | | | | | | KernelScale: 2.743 | | 57 | Accept | 0.18154 | 5.8294 | 0.089231 | 0.10947 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 248 | | | | | | | | | MinLeafSize: 117 | | 58 | Accept | 0.37179 | 0.087804 | 0.089231 | 0.10947 | svm | BoxConstraint: 7.5785 | | | | | | | | | KernelScale: 0.0066815 | | 59 | Accept | 0.37179 | 0.083397 | 0.089231 | 0.10947 | svm | BoxConstraint: 0.0017765 | | | | | | | | | KernelScale: 0.86786 | | 60 | Accept | 0.37179 | 0.081802 | 0.089231 | 0.10947 | svm | BoxConstraint: 0.011465 | | | | | | | | | KernelScale: 0.02747 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 61 | Accept | 0.11692 | 0.070203 | 0.089231 | 0.10947 | tree | MinLeafSize: 12 | | 62 | Accept | 0.29167 | 0.087616 | 0.089231 | 0.10947 | svm | BoxConstraint: 11.939 | | | | | | | | | KernelScale: 11.002 | | 63 | Accept | 0.21795 | 0.067296 | 0.089231 | 0.10947 | knn | NumNeighbors: 6 | | 64 | Accept | 0.18269 | 0.062677 | 0.089231 | 0.10947 | knn | NumNeighbors: 3 | | 65 | Accept | 0.12923 | 0.062417 | 0.089231 | 0.10947 | tree | MinLeafSize: 3 | | 66 | Accept | 0.16923 | 0.066062 | 0.089231 | 0.10947 | tree | MinLeafSize: 56 | | 67 | Accept | 0.1891 | 0.072064 | 0.089231 | 0.10947 | knn | NumNeighbors: 1 | | 68 | Accept | 0.13231 | 10.405 | 0.089231 | 0.10927 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 270 | | | | | | | | | MinLeafSize: 4 | | 69 | Accept | 0.22769 | 0.058514 | 0.089231 | 0.10927 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 70 | Accept | 0.37231 | 0.16623 | 0.089231 | 0.10927 | nb | DistributionNames: kernel | | | | | | | | | Width: 1629.5 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 71 | Accept | 0.16923 | 0.059987 | 0.089231 | 0.10927 | tree | MinLeafSize: 61 | | 72 | Accept | 0.22769 | 0.051395 | 0.089231 | 0.10927 | nb | DistributionNames: normal | | | | | | | | | Width: NaN | | 73 | Accept | 0.12308 | 5.0871 | 0.089231 | 0.10454 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 217 | | | | | | | | | MinLeafSize: 70 | | 74 | Accept | 0.13231 | 10.737 | 0.089231 | 0.10577 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 257 | | | | | | | | | MinLeafSize: 2 | | 75 | Accept | 0.21474 | 0.080252 | 0.089231 | 0.10577 | knn | NumNeighbors: 49 | | 76 | Accept | 0.10154 | 7.3068 | 0.089231 | 0.099465 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 285 | | | | | | | | | MinLeafSize: 2 | | 77 | Accept | 0.12923 | 5.5706 | 0.089231 | 0.10383 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 214 | | | | | | | | | MinLeafSize: 72 | | 78 | Accept | 0.12308 | 5.0042 | 0.089231 | 0.099898 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | MinLeafSize: 73 | | 79 | Accept | 0.13538 | 6.7572 | 0.089231 | 0.1013 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 291 | | | | | | | | | MinLeafSize: 92 | | 80 | Accept | 0.12 | 5.1409 | 0.089231 | 0.099163 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 222 | | | | | | | | | MinLeafSize: 28 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 81 | Accept | 0.14462 | 4.897 | 0.089231 | 0.10034 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 210 | | | | | | | | | MinLeafSize: 102 | | 82 | Best | 0.086154 | 5.009 | 0.086154 | 0.094283 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 208 | | | | | | | | | MinLeafSize: 15 | | 83 | Accept | 0.089231 | 5.6605 | 0.086154 | 0.092125 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 237 | | | | | | | | | MinLeafSize: 15 | | 84 | Accept | 0.092308 | 5.9891 | 0.086154 | 0.092405 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 256 | | | | | | | | | MinLeafSize: 20 | | 85 | Accept | 0.098462 | 4.9373 | 0.086154 | 0.09151 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 202 | | | | | | | | | MinLeafSize: 16 | | 86 | Accept | 0.092308 | 6.354 | 0.086154 | 0.091482 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 262 | | | | | | | | | MinLeafSize: 13 | | 87 | Accept | 0.13846 | 5.1929 | 0.086154 | 0.090767 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 224 | | | | | | | | | MinLeafSize: 98 | | 88 | Accept | 0.10769 | 5.2709 | 0.086154 | 0.093092 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 216 | | | | | | | | | MinLeafSize: 7 | | 89 | Accept | 0.11385 | 6.2613 | 0.086154 | 0.092237 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 263 | | | | | | | | | MinLeafSize: 23 | | 90 | Accept | 0.089231 | 5.1591 | 0.086154 | 0.090261 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 212 | | | | | | | | | MinLeafSize: 15 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 91 | Accept | 0.098462 | 5.5561 | 0.086154 | 0.091164 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 231 | | | | | | | | | MinLeafSize: 14 | | 92 | Accept | 0.092308 | 7.0841 | 0.086154 | 0.090418 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 293 | | | | | | | | | MinLeafSize: 15 | | 93 | Accept | 0.10462 | 5.4911 | 0.086154 | 0.090933 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 200 | | | | | | | | | MinLeafSize: 10 | | 94 | Accept | 0.095385 | 6.2488 | 0.086154 | 0.089667 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 238 | | | | | | | | | MinLeafSize: 20 | | 95 | Accept | 0.14154 | 5.9513 | 0.086154 | 0.090672 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 211 | | | | | | | | | MinLeafSize: 101 | | 96 | Accept | 0.086154 | 7.211 | 0.086154 | 0.089308 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 207 | | | | | | | | | MinLeafSize: 15 | | 97 | Accept | 0.095385 | 7.0247 | 0.086154 | 0.088701 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 220 | | | | | | | | | MinLeafSize: 15 | | 98 | Accept | 0.092308 | 5.6392 | 0.086154 | 0.088919 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 20 | | 99 | Accept | 0.092308 | 5.3506 | 0.086154 | 0.088952 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 207 | | | | | | | | | MinLeafSize: 14 | | 100 | Accept | 0.089231 | 5.0422 | 0.086154 | 0.089122 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 202 | | | | | | | | | MinLeafSize: 15 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 101 | Accept | 0.11077 | 6.5738 | 0.086154 | 0.088935 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 262 | | | | | | | | | MinLeafSize: 23 | | 102 | Accept | 0.092308 | 6.6426 | 0.086154 | 0.088627 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 272 | | | | | | | | | MinLeafSize: 15 | | 103 | Accept | 0.14462 | 5.156 | 0.086154 | 0.088746 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 218 | | | | | | | | | MinLeafSize: 108 | | 104 | Accept | 0.092308 | 7.2137 | 0.086154 | 0.088254 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 295 | | | | | | | | | MinLeafSize: 15 | | 105 | Accept | 0.095385 | 5.2094 | 0.086154 | 0.088642 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 11 | | 106 | Accept | 0.086154 | 5.167 | 0.086154 | 0.088732 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 15 | | 107 | Accept | 0.086154 | 5.0251 | 0.086154 | 0.087649 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 15 | | 108 | Accept | 0.092308 | 4.8932 | 0.086154 | 0.087563 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 201 | | | | | | | | | MinLeafSize: 14 | | 109 | Accept | 0.098462 | 7.0345 | 0.086154 | 0.087822 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 288 | | | | | | | | | MinLeafSize: 20 | | 110 | Accept | 0.10769 | 5.0962 | 0.086154 | 0.087788 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 10 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 111 | Accept | 0.092308 | 5.0295 | 0.086154 | 0.087969 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 206 | | | | | | | | | MinLeafSize: 20 | | 112 | Accept | 0.14769 | 5.0148 | 0.086154 | 0.088051 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 214 | | | | | | | | | MinLeafSize: 110 | | 113 | Accept | 0.086154 | 4.9943 | 0.086154 | 0.087118 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 201 | | | | | | | | | MinLeafSize: 15 | | 114 | Accept | 0.14154 | 6.1042 | 0.086154 | 0.087477 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 238 | | | | | | | | | MinLeafSize: 110 | | 115 | Accept | 0.098462 | 5.589 | 0.086154 | 0.087375 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 217 | | | | | | | | | MinLeafSize: 15 | | 116 | Accept | 0.086154 | 5.0709 | 0.086154 | 0.087064 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 15 | | 117 | Accept | 0.086154 | 5.0476 | 0.086154 | 0.086927 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 203 | | | | | | | | | MinLeafSize: 15 | | 118 | Accept | 0.095385 | 5.1616 | 0.086154 | 0.086424 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 206 | | | | | | | | | MinLeafSize: 11 | | 119 | Accept | 0.17231 | 7.5631 | 0.086154 | 0.086811 | ensemble | Method: Bag | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 104 | | 120 | Accept | 0.11385 | 6.1048 | 0.086154 | 0.086479 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 256 | | | | | | | | | MinLeafSize: 26 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 121 | Accept | 0.10154 | 4.9144 | 0.086154 | 0.086731 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 16 | | 122 | Accept | 0.11385 | 6.6916 | 0.086154 | 0.086764 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 290 | | | | | | | | | MinLeafSize: 26 | | 123 | Accept | 0.17846 | 4.9356 | 0.086154 | 0.086982 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | MinLeafSize: 121 | | 124 | Accept | 0.18154 | 6.7496 | 0.086154 | 0.086624 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 295 | | | | | | | | | MinLeafSize: 124 | | 125 | Accept | 0.092308 | 4.856 | 0.086154 | 0.086795 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 207 | | | | | | | | | MinLeafSize: 46 | | 126 | Accept | 0.086154 | 4.8119 | 0.086154 | 0.086787 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 203 | | | | | | | | | MinLeafSize: 15 | | 127 | Accept | 0.13231 | 4.7345 | 0.086154 | 0.086511 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 58 | | 128 | Accept | 0.15385 | 7.0641 | 0.086154 | 0.086611 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 299 | | | | | | | | | MinLeafSize: 112 | | 129 | Accept | 0.086154 | 5.0542 | 0.086154 | 0.08658 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 207 | | | | | | | | | MinLeafSize: 15 | | 130 | Accept | 0.2 | 4.7512 | 0.086154 | 0.086476 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 201 | | | | | | | | | MinLeafSize: 126 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 131 | Accept | 0.13231 | 5.0661 | 0.086154 | 0.086354 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 53 | | 132 | Accept | 0.089231 | 5.0199 | 0.086154 | 0.086357 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 15 | | 133 | Accept | 0.098462 | 4.7946 | 0.086154 | 0.086579 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 204 | | | | | | | | | MinLeafSize: 48 | | 134 | Accept | 0.092308 | 4.8843 | 0.086154 | 0.086467 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 50 | | 135 | Best | 0.083077 | 4.9469 | 0.083077 | 0.085907 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 205 | | | | | | | | | MinLeafSize: 15 | | 136 | Accept | 0.095385 | 4.954 | 0.083077 | 0.085641 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 210 | | | | | | | | | MinLeafSize: 41 | | 137 | Accept | 0.098462 | 4.9278 | 0.083077 | 0.085638 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 210 | | | | | | | | | MinLeafSize: 51 | | 138 | Accept | 0.10154 | 4.9427 | 0.083077 | 0.085538 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 209 | | | | | | | | | MinLeafSize: 51 | | 139 | Accept | 0.089231 | 5.0365 | 0.083077 | 0.085955 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 215 | | | | | | | | | MinLeafSize: 45 | | 140 | Accept | 0.083077 | 4.8738 | 0.083077 | 0.085392 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 205 | | | | | | | | | MinLeafSize: 15 | |=================================================================================================================================| | Iter | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | result | loss | & validation (sec)| validation loss | validation loss | | | |=================================================================================================================================| | 141 | Accept | 0.10769 | 5.0619 | 0.083077 | 0.085017 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | MinLeafSize: 23 | | 142 | Accept | 0.089231 | 5.1143 | 0.083077 | 0.085054 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 215 | | | | | | | | | MinLeafSize: 49 | | 143 | Accept | 0.089231 | 5.5325 | 0.083077 | 0.085025 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 237 | | | | | | | | | MinLeafSize: 49 | | 144 | Accept | 0.089231 | 5.4298 | 0.083077 | 0.084846 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 234 | | | | | | | | | MinLeafSize: 49 | | 145 | Accept | 0.092308 | 5.4706 | 0.083077 | 0.084781 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 224 | | | | | | | | | MinLeafSize: 43 | | 146 | Accept | 0.098462 | 6.0056 | 0.083077 | 0.084933 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 248 | | | | | | | | | MinLeafSize: 51 | | 147 | Accept | 0.086154 | 6.3296 | 0.083077 | 0.084906 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 263 | | | | | | | | | MinLeafSize: 49 | | 148 | Accept | 0.083077 | 6.4813 | 0.083077 | 0.084823 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 273 | | | | | | | | | MinLeafSize: 49 | | 149 | Accept | 0.086154 | 7.0564 | 0.083077 | 0.085268 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 298 | | | | | | | | | MinLeafSize: 49 | | 150 | Accept | 0.095385 | 6.4287 | 0.083077 | 0.085022 | ensemble | Method: LogitBoost | | | | | | | | | NumLearningCycles: 268 | | | | | | | | | MinLeafSize: 37 |
__________________________________________________________ Optimization completed. Total iterations: 150 Total elapsed time: 698.3513 seconds Total time for training and validation: 553.6354 seconds Best observed learner is an ensemble model with: Method: LogitBoost NumLearningCycles: 205 MinLeafSize: 15 Observed validation loss: 0.083077 Time for training and validation: 4.9469 seconds Best estimated learner (returned model) is an ensemble model with: Method: LogitBoost NumLearningCycles: 205 MinLeafSize: 15 Estimated validation loss: 0.085022 Estimated time for training and validation: 4.9757 seconds Documentation for fitcauto display
Итоговая модель возвращена fitcauto соответствует лучшему предполагаемому ученику. Прежде, чем возвратить модель, функция переобучает его с помощью целых обучающих данных (carsTrain), перечисленный Learner (или модель) тип и отображенные гиперзначения параметров.
Оцените эффективность набора тестов
Оцените эффективность модели на наборе тестов.
testAccuracy = 1 - loss(Mdl,carsTest,"Origin")testAccuracy = 0.9263
confusionchart(carsTest.Origin,predict(Mdl,carsTest))
Используйте fitcauto автоматически выбрать модель классификации оптимизированными гиперпараметрами, учитывая предиктор и данные об ответе, хранимые в отдельных переменных.
Загрузка данных
Загрузите humanactivity набор данных. Этот набор данных содержит 24 075 наблюдений за пятью физической деятельностью человека: Находясь (1), Стоя (2), Идя (3), Запускаясь (4), и Танцуя (5). Каждое наблюдение имеет 60 функций, извлеченных из ускоряющих данных, измеренных датчиками акселерометра смартфона. Переменная feat содержит матрицу данных предиктора 60 функций этих 24 075 наблюдений и переменную отклика actid содержит идентификаторы действия для наблюдений как целые числа.
load humanactivityДанные о разделе
Разделите данные в наборы обучающих данных и наборы тестов. Используйте 90% наблюдений, чтобы выбрать модель, и 10% наблюдений, чтобы подтвердить итоговую модель, возвращенную fitcauto. Используйте cvpartition зарезервировать 10% наблюдений для тестирования.
rng("default") % For reproducibility of the partition c = cvpartition(actid,"Holdout",0.10); trainingIndices = training(c); % Indices for the training set XTrain = feat(trainingIndices,:); YTrain = actid(trainingIndices); testIndices = test(c); % Indices for the test set XTest = feat(testIndices,:); YTest = actid(testIndices);
Запустите fitcauto
Передайте обучающие данные fitcauto. Поскольку обучающие данные XTrain имеет больше чем 10 000 наблюдений, используйте оптимизацию ASHA, а не Байесовую оптимизацию. fitcauto функционируйте случайным образом выбирает соответствующую модель (или ученик) типы с различными гиперзначениями параметров, обучает модели на небольшом подмножестве обучающих данных, продвигает модели, которые выполняют хорошо, и переобучает продвинутые модели на прогрессивно больших наборах обучающих данных. Функция возвращает модель с лучшей эффективностью перекрестной проверки, переобученной на всех обучающих данных и таблице, которая содержит детали оптимизации. Задайте, чтобы запуститься, оптимизация параллельно (требует Parallel Computing Toolbox™).
По умолчанию, fitcauto предоставляет график оптимизации и итеративное отображение результатов оптимизации. Для получения дополнительной информации о том, как интерпретировать эти результаты, смотрите Многословное Отображение.
options = struct("Optimizer","asha","UseParallel",true); [Mdl,OptimizationResults] = fitcauto(XTrain,YTrain,"HyperparameterOptimizationOptions",options);
Warning: It is recommended that you first standardize all numeric predictors when optimizing the Naive Bayes 'Width' parameter. Ignore this warning if you have done that.
Starting parallel pool (parpool) using the 'local' profile ... Connected to the parallel pool (number of workers: 6). Copying objective function to workers... Done copying objective function to workers. Learner types to explore: ensemble, knn, nb, svm, tree Total iterations (MaxObjectiveEvaluations): 510 Total time (MaxTime): Inf |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 1 | 6 | Best | 0.74165 | 7.0787 | 0.74165 | 271 | tree | MinLeafSize: 723 | | 2 | 2 | Accept | 0.1169 | 9.5275 | 0.043705 | 271 | knn | NumNeighbors: 42 | | 3 | 2 | Accept | 0.74165 | 11.543 | 0.043705 | 271 | knn | NumNeighbors: 1726 | | 4 | 2 | Accept | 0.084872 | 9.3569 | 0.043705 | 271 | knn | NumNeighbors: 27 | | 5 | 2 | Best | 0.043705 | 7.027 | 0.043705 | 271 | tree | MinLeafSize: 7 | | 6 | 2 | Accept | 0.060642 | 7.0193 | 0.043705 | 271 | tree | MinLeafSize: 11 | | 7 | 4 | Accept | 0.044443 | 1.4788 | 0.043705 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 8 | 5 | Accept | 0.058243 | 0.79831 | 0.028983 | 271 | tree | MinLeafSize: 25 | | 9 | 5 | Best | 0.028983 | 2.1507 | 0.028983 | 1084 | tree | MinLeafSize: 7 | | 10 | 6 | Accept | 0.049566 | 1.0799 | 0.028983 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 11 | 6 | Accept | 0.74165 | 5.5337 | 0.028983 | 271 | knn | NumNeighbors: 3072 | | 12 | 6 | Accept | 0.74155 | 5.2384 | 0.028983 | 271 | knn | NumNeighbors: 178 | | 13 | 6 | Accept | 0.53974 | 5.9615 | 0.028983 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.44923 | | | | | | | | | | KernelScale: 517.98 | | 14 | 6 | Accept | 0.74165 | 5.9452 | 0.028983 | 271 | knn | NumNeighbors: 294 | | 15 | 6 | Accept | 0.038259 | 1.6449 | 0.028983 | 1084 | tree | MinLeafSize: 25 | | 16 | 6 | Accept | 0.11579 | 17.724 | 0.028983 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 218 | | | | | | | | | | MinLeafSize: 60 | | | | | | | | | | MaxNumSplits: 12 | | 17 | 5 | Accept | 0.73966 | 12.951 | 0.028983 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 585.98 | | | | | | | | | | KernelScale: 0.021825 | | 18 | 5 | Accept | 0.68617 | 12.504 | 0.028983 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0058216 | | | | | | | | | | KernelScale: 0.45032 | | 19 | 6 | Accept | 0.74165 | 11.499 | 0.028983 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.0634 | | | | | | | | | | KernelScale: 0.0035173 | | 20 | 6 | Accept | 0.035582 | 1.612 | 0.028983 | 1084 | tree | MinLeafSize: 11 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 21 | 6 | Accept | 0.74008 | 3.8698 | 0.028983 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.040932 | | | | | | | | | | KernelScale: 949.09 | | 22 | 6 | Best | 0.02326 | 1.302 | 0.02326 | 4334 | tree | MinLeafSize: 7 | | 23 | 6 | Accept | 0.7057 | 12.734 | 0.02326 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 53.678 | | | | | | | | | | KernelScale: 0.3797 | | 24 | 6 | Accept | 0.055289 | 2.0309 | 0.02326 | 271 | knn | NumNeighbors: 5 | | 25 | 6 | Accept | 0.051966 | 2.2552 | 0.02326 | 271 | knn | NumNeighbors: 7 | | 26 | 6 | Accept | 0.74165 | 6.4874 | 0.02326 | 271 | knn | NumNeighbors: 1240 | | 27 | 6 | Accept | 0.73962 | 12.459 | 0.02326 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.099738 | | | | | | | | | | KernelScale: 0.0018954 | | 28 | 6 | Accept | 0.1439 | 3.0955 | 0.02326 | 271 | knn | NumNeighbors: 44 | | 29 | 6 | Accept | 0.045874 | 2.1894 | 0.02326 | 271 | knn | NumNeighbors: 2 | | 30 | 6 | Accept | 0.035306 | 15.731 | 0.02326 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | | MinLeafSize: 12 | | | | | | | | | | MaxNumSplits: 65 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 31 | 6 | Accept | 0.043197 | 7.5214 | 0.02326 | 1084 | knn | NumNeighbors: 7 | | 32 | 6 | Accept | 0.055381 | 1.9818 | 0.02326 | 271 | knn | NumNeighbors: 3 | | 33 | 6 | Accept | 0.054181 | 0.57186 | 0.02326 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 34 | 6 | Accept | 0.74165 | 0.21834 | 0.02326 | 271 | tree | MinLeafSize: 369 | | 35 | 6 | Accept | 0.74165 | 5.8009 | 0.02326 | 271 | knn | NumNeighbors: 1193 | | 36 | 6 | Accept | 0.033921 | 6.9268 | 0.02326 | 1084 | knn | NumNeighbors: 2 | | 37 | 6 | Accept | 0.057089 | 1.2849 | 0.02326 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 38 | 6 | Accept | 0.35559 | 2.2316 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 3.2454 | | | | | | | | | | KernelScale: 500.94 | | 39 | 6 | Accept | 0.047536 | 2.6934 | 0.02326 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 432.63 | | | | | | | | | | KernelScale: 373.97 | | 40 | 6 | Accept | 0.032075 | 2.0195 | 0.02326 | 1084 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 432.63 | | | | | | | | | | KernelScale: 373.97 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 41 | 6 | Accept | 0.10859 | 2.0835 | 0.02326 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0052814 | | | | | | | | | | KernelScale: 546.04 | | 42 | 6 | Accept | 0.74165 | 7.2993 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.0068861 | | | | | | | | | | KernelScale: 1.9477 | | 43 | 6 | Accept | 0.74165 | 5.6881 | 0.02326 | 271 | knn | NumNeighbors: 524 | | 44 | 6 | Accept | 0.051089 | 2.0414 | 0.02326 | 271 | knn | NumNeighbors: 3 | | 45 | 6 | Accept | 0.7398 | 10.815 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 5.009 | | | | | | | | | | KernelScale: 0.0070102 | | 46 | 6 | Accept | 0.03069 | 25.357 | 0.02326 | 1084 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | | MinLeafSize: 12 | | | | | | | | | | MaxNumSplits: 65 | | 47 | 6 | Accept | 0.036552 | 7.16 | 0.02326 | 1084 | knn | NumNeighbors: 3 | | 48 | 6 | Accept | 0.047166 | 1.8295 | 0.02326 | 271 | knn | NumNeighbors: 2 | | 49 | 6 | Accept | 0.049105 | 0.49734 | 0.02326 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 50 | 6 | Accept | 0.73934 | 10.403 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 198.87 | | | | | | | | | | KernelScale: 0.085112 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 51 | 6 | Accept | 0.7314 | 74.726 | 0.02326 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 9.8268e-06 | | 52 | 6 | Accept | 0.74165 | 10.108 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.24903 | | | | | | | | | | KernelScale: 0.043685 | | 53 | 6 | Accept | 0.036413 | 6.719 | 0.02326 | 1084 | knn | NumNeighbors: 2 | | 54 | 6 | Accept | 0.060781 | 2.3593 | 0.02326 | 271 | knn | NumNeighbors: 13 | | 55 | 6 | Accept | 0.74063 | 11.248 | 0.02326 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.51791 | | | | | | | | | | KernelScale: 0.32403 | | 56 | 6 | Accept | 0.73029 | 62.941 | 0.02326 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 5.0726e-11 | | 57 | 6 | Accept | 0.051966 | 0.86077 | 0.02326 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 58 | 6 | Accept | 0.052197 | 1.8288 | 0.02326 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 2.8638 | | | | | | | | | | KernelScale: 201.68 | | 59 | 6 | Accept | 0.74165 | 11.549 | 0.02326 | 271 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 203 | | | | | | | | | | MinLeafSize: 3190 | | | | | | | | | | MaxNumSplits: 11 | | 60 | 6 | Accept | 0.74165 | 6.704 | 0.02326 | 271 | knn | NumNeighbors: 2890 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 61 | 6 | Accept | 0.74165 | 5.8589 | 0.02326 | 271 | knn | NumNeighbors: 762 | | 62 | 6 | Accept | 0.1133 | 3.0194 | 0.02326 | 271 | knn | NumNeighbors: 46 | | 63 | 6 | Accept | 0.049151 | 0.53626 | 0.02326 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 64 | 6 | Accept | 0.041767 | 4.6298 | 0.02326 | 1084 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 2.8638 | | | | | | | | | | KernelScale: 201.68 | | 65 | 6 | Accept | 0.74165 | 5.894 | 0.02326 | 271 | knn | NumNeighbors: 822 | | 66 | 6 | Accept | 0.056581 | 1.3414 | 0.02326 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 67 | 6 | Accept | 0.74165 | 5.5647 | 0.02326 | 271 | knn | NumNeighbors: 279 | | 68 | 6 | Accept | 0.15419 | 3.0206 | 0.02326 | 271 | knn | NumNeighbors: 54 | | 69 | 6 | Accept | 0.024691 | 45.811 | 0.02326 | 4334 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 213 | | | | | | | | | | MinLeafSize: 12 | | | | | | | | | | MaxNumSplits: 65 | | 70 | 6 | Best | 0.022614 | 11.185 | 0.022614 | 4334 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 432.63 | | | | | | | | | | KernelScale: 373.97 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 71 | 6 | Accept | 0.74165 | 9.802 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.0010658 | | | | | | | | | | KernelScale: 0.1049 | | 72 | 6 | Accept | 0.73902 | 10.995 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 32.297 | | | | | | | | | | KernelScale: 0.04955 | | 73 | 6 | Accept | 0.049797 | 0.58503 | 0.022614 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 74 | 6 | Accept | 0.74165 | 13.644 | 0.022614 | 271 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 254 | | | | | | | | | | MinLeafSize: 8596 | | | | | | | | | | MaxNumSplits: 73 | | 75 | 6 | Accept | 0.74165 | 11.107 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.23597 | | | | | | | | | | KernelScale: 0.0011739 | | 76 | 6 | Accept | 0.048274 | 1.8954 | 0.022614 | 271 | knn | NumNeighbors: 2 | | 77 | 6 | Accept | 0.099903 | 2.4157 | 0.022614 | 271 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 223 | | | | | | | | | | MinLeafSize: 1 | | | | | | | | | | MaxNumSplits: 75 | | 78 | 6 | Accept | 0.74165 | 8.7174 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.13626 | | | | | | | | | | KernelScale: 0.37002 | | 79 | 6 | Accept | 0.035444 | 6.7619 | 0.022614 | 1084 | knn | NumNeighbors: 2 | | 80 | 6 | Accept | 0.73943 | 11.389 | 0.022614 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 6.476 | | | | | | | | | | KernelScale: 0.024305 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 81 | 6 | Accept | 0.74165 | 0.16927 | 0.022614 | 271 | tree | MinLeafSize: 192 | | 82 | 6 | Accept | 0.28443 | 6.1819 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 4.6764 | | | | | | | | | | KernelScale: 2.4604 | | 83 | 6 | Accept | 0.74165 | 5.5328 | 0.022614 | 271 | knn | NumNeighbors: 5216 | | 84 | 6 | Accept | 0.73994 | 10.148 | 0.022614 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.64248 | | | | | | | | | | KernelScale: 0.011614 | | 85 | 6 | Accept | 0.74165 | 5.4912 | 0.022614 | 271 | knn | NumNeighbors: 5719 | | 86 | 6 | Accept | 0.064658 | 0.56457 | 0.022614 | 271 | tree | MinLeafSize: 35 | | 87 | 6 | Accept | 0.038398 | 6.9574 | 0.022614 | 1084 | knn | NumNeighbors: 5 | | 88 | 6 | Accept | 0.034475 | 6.8188 | 0.022614 | 1084 | knn | NumNeighbors: 3 | | 89 | 6 | Accept | 0.74165 | 0.14527 | 0.022614 | 271 | tree | MinLeafSize: 2993 | | 90 | 6 | Accept | 0.058058 | 0.46957 | 0.022614 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 91 | 6 | Accept | 0.026629 | 22.134 | 0.022614 | 4334 | knn | NumNeighbors: 2 | | 92 | 6 | Accept | 0.74165 | 15.612 | 0.022614 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 278 | | | | | | | | | | MinLeafSize: 3475 | | | | | | | | | | MaxNumSplits: 38 | | 93 | 6 | Accept | 0.056258 | 0.62655 | 0.022614 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 94 | 6 | Accept | 0.050858 | 0.51896 | 0.022614 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 95 | 6 | Accept | 0.74165 | 5.8513 | 0.022614 | 271 | knn | NumNeighbors: 810 | | 96 | 6 | Accept | 0.061796 | 0.58768 | 0.022614 | 271 | tree | MinLeafSize: 35 | | 97 | 6 | Accept | 0.73002 | 62.357 | 0.022614 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 1.7322e-07 | | 98 | 6 | Accept | 0.064011 | 215.57 | 0.022614 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 2.3959 | | 99 | 6 | Accept | 0.046659 | 0.93439 | 0.022614 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 100 | 6 | Accept | 0.052889 | 0.48086 | 0.022614 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 101 | 6 | Accept | 0.10033 | 2.0363 | 0.022614 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0041303 | | | | | | | | | | KernelScale: 185.74 | | 102 | 6 | Accept | 0.045782 | 0.48207 | 0.022614 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 103 | 6 | Accept | 0.042874 | 0.7858 | 0.022614 | 271 | tree | MinLeafSize: 1 | | 104 | 6 | Accept | 0.029167 | 1.4546 | 0.022614 | 1084 | tree | MinLeafSize: 1 | | 105 | 6 | Best | 0.020537 | 1.3072 | 0.020537 | 4334 | tree | MinLeafSize: 1 | | 106 | 6 | Accept | 0.41033 | 211.85 | 0.020537 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 68.422 | | 107 | 6 | Accept | 0.74165 | 5.4562 | 0.020537 | 271 | knn | NumNeighbors: 7694 | | 108 | 6 | Accept | 0.74165 | 0.29134 | 0.020537 | 271 | tree | MinLeafSize: 675 | | 109 | 5 | Accept | 0.72979 | 56.027 | 0.020537 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 7.594e-10 | | 110 | 5 | Accept | 0.047212 | 1.914 | 0.020537 | 271 | knn | NumNeighbors: 5 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 111 | 6 | Accept | 0.049982 | 0.50009 | 0.020537 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 112 | 6 | Accept | 0.051781 | 0.56465 | 0.020537 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 113 | 6 | Accept | 0.061658 | 0.50308 | 0.020537 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 114 | 6 | Accept | 0.73851 | 9.9938 | 0.020537 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 1.0952 | | | | | | | | | | KernelScale: 0.0011239 | | 115 | 6 | Best | 0.015507 | 143.46 | 0.015507 | 17335 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 432.63 | | | | | | | | | | KernelScale: 373.97 | | 116 | 6 | Accept | 0.038951 | 7.502 | 0.015507 | 1084 | knn | NumNeighbors: 5 | | 117 | 6 | Accept | 0.058658 | 2.2751 | 0.015507 | 271 | knn | NumNeighbors: 9 | | 118 | 6 | Accept | 0.56535 | 210.03 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 175.22 | | 119 | 6 | Accept | 0.050628 | 1.9935 | 0.015507 | 271 | knn | NumNeighbors: 4 | | 120 | 6 | Accept | 0.73126 | 9.0011 | 0.015507 | 271 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 230.2 | | | | | | | | | | KernelScale: 0.32321 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 121 | 6 | Accept | 0.51278 | 8.8728 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.011209 | | | | | | | | | | KernelScale: 1.0514 | | 122 | 6 | Accept | 0.037752 | 6.7081 | 0.015507 | 1084 | knn | NumNeighbors: 4 | | 123 | 6 | Accept | 0.74165 | 204.7 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 1679.5 | | 124 | 6 | Accept | 0.046751 | 1.8383 | 0.015507 | 271 | knn | NumNeighbors: 2 | | 125 | 6 | Accept | 0.73108 | 58.241 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 7.7951e-15 | | 126 | 6 | Accept | 0.74165 | 12.265 | 0.015507 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 216 | | | | | | | | | | MinLeafSize: 421 | | | | | | | | | | MaxNumSplits: 59 | | 127 | 6 | Accept | 0.048182 | 0.46464 | 0.015507 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 128 | 6 | Accept | 0.078687 | 0.736 | 0.015507 | 271 | tree | MinLeafSize: 12 | | 129 | 6 | Accept | 0.034429 | 6.89 | 0.015507 | 1084 | knn | NumNeighbors: 2 | | 130 | 6 | Accept | 0.74165 | 5.1529 | 0.015507 | 271 | knn | NumNeighbors: 893 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 131 | 6 | Accept | 0.04832 | 0.42408 | 0.015507 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 132 | 6 | Accept | 0.045828 | 0.40531 | 0.015507 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 133 | 6 | Accept | 0.74165 | 0.11469 | 0.015507 | 271 | tree | MinLeafSize: 943 | | 134 | 6 | Accept | 0.73163 | 58.558 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 4.0016e-08 | | 135 | 6 | Accept | 0.11732 | 0.47758 | 0.015507 | 271 | tree | MinLeafSize: 43 | | 136 | 6 | Accept | 0.051181 | 0.45611 | 0.015507 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 137 | 6 | Accept | 0.50757 | 116.52 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.005237 | | 138 | 6 | Accept | 0.74165 | 5.0747 | 0.015507 | 271 | knn | NumNeighbors: 2792 | | 139 | 6 | Accept | 0.74165 | 5.0727 | 0.015507 | 271 | knn | NumNeighbors: 1886 | | 140 | 6 | Accept | 0.27972 | 6.2223 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0020942 | | | | | | | | | | KernelScale: 1.6249 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 141 | 6 | Accept | 0.04652 | 0.41269 | 0.015507 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 142 | 6 | Accept | 0.11565 | 0.45669 | 0.015507 | 271 | tree | MinLeafSize: 46 | | 143 | 6 | Accept | 0.045736 | 0.65924 | 0.015507 | 271 | tree | MinLeafSize: 2 | | 144 | 6 | Accept | 0.11621 | 0.39888 | 0.015507 | 271 | tree | MinLeafSize: 59 | | 145 | 6 | Accept | 0.026721 | 20.503 | 0.015507 | 4334 | knn | NumNeighbors: 2 | | 146 | 6 | Accept | 0.29744 | 7.5199 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 318.91 | | | | | | | | | | KernelScale: 1.4414 | | 147 | 6 | Accept | 0.026814 | 1.3329 | 0.015507 | 1084 | tree | MinLeafSize: 2 | | 148 | 6 | Accept | 0.020537 | 1.1856 | 0.015507 | 4334 | tree | MinLeafSize: 2 | | 149 | 6 | Accept | 0.74165 | 5.0812 | 0.015507 | 271 | knn | NumNeighbors: 9834 | | 150 | 6 | Accept | 0.74165 | 4.9886 | 0.015507 | 271 | knn | NumNeighbors: 3626 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 151 | 6 | Accept | 0.10638 | 1.9037 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.35259 | | | | | | | | | | KernelScale: 56.147 | | 152 | 6 | Accept | 0.049012 | 0.70414 | 0.015507 | 271 | tree | MinLeafSize: 1 | | 153 | 6 | Accept | 0.023537 | 1.4334 | 0.015507 | 1084 | tree | MinLeafSize: 1 | | 154 | 5 | Accept | 0.73011 | 60.803 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 1.8217e-05 | | 155 | 5 | Accept | 0.73486 | 8.5215 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0023583 | | | | | | | | | | KernelScale: 0.24236 | | 156 | 6 | Accept | 0.047074 | 1.0388 | 0.015507 | 271 | tree | MinLeafSize: 1 | | 157 | 6 | Accept | 0.70237 | 8.6088 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.062538 | | | | | | | | | | KernelScale: 0.37767 | | 158 | 6 | Accept | 0.74165 | 4.9997 | 0.015507 | 271 | knn | NumNeighbors: 323 | | 159 | 6 | Accept | 0.029675 | 1.4729 | 0.015507 | 1084 | tree | MinLeafSize: 1 | | 160 | 6 | Accept | 0.74165 | 9.302 | 0.015507 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 214 | | | | | | | | | | MinLeafSize: 775 | | | | | | | | | | MaxNumSplits: 50 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 161 | 6 | Accept | 0.37119 | 3.7119 | 0.015507 | 271 | knn | NumNeighbors: 115 | | 162 | 6 | Accept | 0.74165 | 5.107 | 0.015507 | 271 | knn | NumNeighbors: 4695 | | 163 | 6 | Accept | 0.048043 | 0.5202 | 0.015507 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 164 | 6 | Accept | 0.044951 | 0.46331 | 0.015507 | 271 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 165 | 6 | Accept | 0.74165 | 12.879 | 0.015507 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 284 | | | | | | | | | | MinLeafSize: 4201 | | | | | | | | | | MaxNumSplits: 68 | | 166 | 6 | Accept | 0.049243 | 1.7758 | 0.015507 | 271 | knn | NumNeighbors: 2 | | 167 | 6 | Accept | 0.59004 | 90.952 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.0015851 | | 168 | 6 | Accept | 0.050858 | 0.47237 | 0.015507 | 1084 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 169 | 6 | Accept | 0.021276 | 1.3237 | 0.015507 | 4334 | tree | MinLeafSize: 1 | | 170 | 6 | Accept | 0.033183 | 19.828 | 0.015507 | 271 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 292 | | | | | | | | | | MinLeafSize: 12 | | | | | | | | | | MaxNumSplits: 85 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 171 | 6 | Accept | 0.11422 | 2.7931 | 0.015507 | 271 | knn | NumNeighbors: 41 | | 172 | 6 | Accept | 0.74165 | 0.10587 | 0.015507 | 271 | tree | MinLeafSize: 708 | | 173 | 6 | Accept | 0.73929 | 10.665 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.44289 | | | | | | | | | | KernelScale: 0.0062786 | | 174 | 6 | Accept | 0.73076 | 53.017 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 1.203e-13 | | 175 | 6 | Accept | 0.058058 | 1.9539 | 0.015507 | 271 | knn | NumNeighbors: 6 | | 176 | 6 | Accept | 0.73076 | 4.041 | 0.015507 | 271 | knn | NumNeighbors: 131 | | 177 | 6 | Accept | 0.01583 | 6.3126 | 0.015507 | 17335 | tree | MinLeafSize: 1 | | 178 | 6 | Accept | 0.05635 | 6.521 | 0.015507 | 271 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 216 | | | | | | | | | | MinLeafSize: 2 | | | | | | | | | | MaxNumSplits: 42 | | 179 | 6 | Accept | 0.74165 | 5.2228 | 0.015507 | 271 | knn | NumNeighbors: 3347 | | 180 | 6 | Accept | 0.74165 | 5.12 | 0.015507 | 271 | knn | NumNeighbors: 4390 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 181 | 6 | Accept | 0.09401 | 2.3294 | 0.015507 | 271 | knn | NumNeighbors: 24 | | 182 | 6 | Accept | 0.27811 | 5.2153 | 0.015507 | 271 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 35.627 | | | | | | | | | | KernelScale: 3.1484 | | 183 | 6 | Accept | 0.07038 | 2.0202 | 0.015507 | 271 | knn | NumNeighbors: 10 | | 184 | 6 | Accept | 0.035167 | 6.3428 | 0.015507 | 1084 | knn | NumNeighbors: 2 | | 185 | 6 | Accept | 0.051874 | 0.73088 | 0.015507 | 271 | tree | MinLeafSize: 3 | | 186 | 6 | Accept | 0.26241 | 0.25171 | 0.015507 | 271 | tree | MinLeafSize: 79 | | 187 | 6 | Accept | 0.74165 | 5.2683 | 0.015507 | 271 | knn | NumNeighbors: 3004 | | 188 | 6 | Accept | 0.7314 | 54.858 | 0.015507 | 271 | nb | DistributionNames: kernel | | | | | | | | | | Width: 7.872e-07 | | 189 | 6 | Accept | 0.030321 | 1.4975 | 0.015507 | 1084 | tree | MinLeafSize: 3 | | 190 | 6 | Accept | 0.74165 | 5.4671 | 0.015507 | 271 | knn | NumNeighbors: 9464 | |========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Training set | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | size | | | |========================================================================================================================================| | 191 | 6 | Accept | 0.74165 | 5.3621 | 0.015507 | 271 | knn | NumNeighbors: 748 | | 192 | 6 | Accept | 0.054735 | 2.0058 | 0.015507 | 271 | knn | NumNeighbors: 6 | | 193 | 6 | Accept | 0.74165 | 11.604 | 0.015507 | 271 | ensemble | Method: AdaBoostM2 | | | | | ...
__________________________________________________________ Optimization completed. Total iterations: 510 Total elapsed time: 1229.4372 seconds Total time for training and validation: 6958.3168 seconds Best observed learner is an ensemble model with: Method: AdaBoostM2 NumLearningCycles: 213 MinLeafSize: 7 MaxNumSplits: 82 Observed validation loss: 0.0035998 Time for training and validation: 83.6534 seconds Documentation for fitcauto display
Итоговая модель возвращена fitcauto соответствует лучшему наблюдаемому ученику. Прежде, чем возвратить модель, функция переобучает его с помощью всех обучающих данных (XTrain и YTrain), перечисленный Learner (или модель) тип и отображенные гиперзначения параметров.
Оцените эффективность набора тестов
Оцените итоговую производительность модели на наборе тестовых данных.
testAccuracy = 1 - loss(Mdl,XTest,YTest)
testAccuracy = 0.9963
Итоговая модель правильно классифицирует более чем 99% наблюдений.
Используйте fitcauto автоматически выбрать модель классификации оптимизированными гиперпараметрами, учитывая предиктор и данные об ответе, хранимые в таблице. Перед передающими данными к fitcauto, выполните выбор признаков, чтобы удалить неважные предикторы из набора данных.
Загрузка и данные о разделе
Считайте файл примера CreditRating_Historical.dat в таблицу. Данные о предикторе состоят из финансовых отношений и информации об отрасли промышленности для списка корпоративных клиентов. Переменная отклика состоит из кредитных рейтингов, присвоенных рейтинговым агентством. Предварительно просмотрите первые несколько строк набора данных.
creditrating = readtable("CreditRating_Historical.dat");
head(creditrating)ans=8×8 table
ID WC_TA RE_TA EBIT_TA MVE_BVTD S_TA Industry Rating
_____ ______ ______ _______ ________ _____ ________ _______
62394 0.013 0.104 0.036 0.447 0.142 3 {'BB' }
48608 0.232 0.335 0.062 1.969 0.281 8 {'A' }
42444 0.311 0.367 0.074 1.935 0.366 1 {'A' }
48631 0.194 0.263 0.062 1.017 0.228 4 {'BBB'}
43768 0.121 0.413 0.057 3.647 0.466 12 {'AAA'}
39255 -0.117 -0.799 0.01 0.179 0.082 4 {'CCC'}
62236 0.087 0.158 0.049 0.816 0.324 2 {'BBB'}
39354 0.005 0.181 0.034 2.597 0.388 7 {'AA' }
Поскольку каждое значение в ID переменная является уникальным идентификатором клиента, то есть, length(unique(creditrating.ID)) равно количеству наблюдений в creditrating, ID переменная является плохим предиктором. Удалите ID переменная из таблицы, и преобразует Industry переменная к categorical переменная.
creditrating = removevars(creditrating,"ID");
creditrating.Industry = categorical(creditrating.Industry);Разделите данные в наборы обучающих данных и наборы тестов. Используйте приблизительно 85% наблюдений для выбора модели и настраивающего процесса гиперпараметра, и 15% наблюдений, чтобы проверить производительность итоговой модели, возвращенной fitcauto на новых данных. Используйте cvpartition разделить данные.
rng("default") % For reproducibility of the partition c = cvpartition(creditrating.Rating,"Holdout",0.15); trainingIndices = training(c); % Indices for the training set testIndices = test(c); % Indices for the test set creditTrain = creditrating(trainingIndices,:); creditTest = creditrating(testIndices,:);
Выполните выбор признаков
Прежде, чем передать обучающие данные fitcauto, найдите важные предикторы при помощи fscchi2 функция. Визуализируйте баллы предиктора при помощи bar функция. Поскольку некоторыми баллами может быть Inf, и bar отбрасывания Inf значения, постройте конечные баллы сначала и затем постройте конечное представление Inf баллы в различном цвете.
[idx,scores] = fscchi2(creditTrain,"Rating"); bar(scores(idx)) % Represents finite scores hold on veryImportant = isinf(scores); finiteMax = max(scores(~veryImportant)); bar(finiteMax*veryImportant(idx)) % Represents Inf scores hold off xticklabels(strrep(creditTrain.Properties.VariableNames(idx),"_","\_")) xtickangle(45) legend(["Finite Scores","Inf Scores"])
Заметьте, что Industry предиктор имеет низкий балл, соответствующий p-значению, которое больше 0.05, который указывает на тот Industry не может быть важная функция. Удалите Industry покажите от наборов тестовых данных и обучения.
creditTrain = removevars(creditTrain,'Industry'); creditTest = removevars(creditTest,'Industry');
Запустите fitcauto
Передайте обучающие данные fitcauto. Функция использует Байесовую оптимизацию, чтобы выбрать модели и их гиперзначения параметров, и возвращает обученную модель Mdl с лучшей ожидаемой эффективностью. Задайте, чтобы попробовать все доступные типы ученика и запуститься, оптимизация параллельно (требует Parallel Computing Toolbox™). Возвратите второй выход Results это содержит детали Байесовой оптимизации.
Ожидайте, что этот процесс займет время. По умолчанию, fitcauto предоставляет график оптимизации и итеративное отображение результатов оптимизации. Для получения дополнительной информации о том, как интерпретировать эти результаты, смотрите Многословное Отображение.
options = struct("UseParallel",true); [Mdl,Results] = fitcauto(creditTrain,"Rating", ... "Learners","all","HyperparameterOptimizationOptions",options);
Warning: It is recommended that you first standardize all numeric predictors when optimizing the Naive Bayes 'Width' parameter. Ignore this warning if you have done that.
Starting parallel pool (parpool) using the 'local' profile ... Connected to the parallel pool (number of workers: 6). Copying objective function to workers... Done copying objective function to workers. Learner types to explore: discr, ensemble, kernel, knn, linear, nb, svm, tree Total iterations (MaxObjectiveEvaluations): 240 Total time (MaxTime): Inf |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 1 | 6 | Best | 0.42716 | 4.9077 | 0.42716 | 0.42716 | discr | Delta: 0.00046441 | | | | | | | | | | Gamma: 0.2485 | | 2 | 4 | Accept | 0.74185 | 7.2984 | 0.24948 | 0.29794 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.48455 | | | | | | | | | | KernelScale: 354.44 | | 3 | 4 | Best | 0.24948 | 7.9291 | 0.24948 | 0.29794 | linear | Coding: onevsone | | | | | | | | | | Lambda: 6.3551e-08 | | | | | | | | | | Learner: logistic | | 4 | 4 | Accept | 0.29794 | 6.1613 | 0.24948 | 0.29794 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 12 | | | | | | | | | | LearnRate: 0.063776 | | | | | | | | | | MinLeafSize: 277 | | 5 | 4 | Accept | 0.25067 | 1.155 | 0.24948 | 0.25067 | knn | NumNeighbors: 105 | | | | | | | | | | Distance: minkowski | | 6 | 6 | Accept | 0.52917 | 2.3222 | 0.24948 | 0.25067 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.002417 | | | | | | | | | | KernelScale: 356.9 | | 7 | 4 | Accept | 0.6904 | 1.5015 | 0.24948 | 0.29794 | knn | NumNeighbors: 255 | | | | | | | | | | Distance: jaccard | | 8 | 4 | Accept | 0.6904 | 1.3877 | 0.24948 | 0.29794 | knn | NumNeighbors: 255 | | | | | | | | | | Distance: jaccard | | 9 | 4 | Accept | 0.6904 | 1.3696 | 0.24948 | 0.29794 | knn | NumNeighbors: 255 | | | | | | | | | | Distance: jaccard | | 10 | 4 | Accept | 0.55818 | 1.0769 | 0.24948 | 0.29794 | discr | Delta: 0.98612 | | | | | | | | | | Gamma: 0.86519 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 11 | 6 | Best | 0.24649 | 17.118 | 0.24649 | 0.24649 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 7.8433 | | | | | | | | | | Lambda: 1.4468e-06 | | 12 | 6 | Accept | 0.26413 | 1.1368 | 0.24649 | 0.24649 | tree | MinLeafSize: 30 | | 13 | 6 | Accept | 0.45169 | 1.5112 | 0.24649 | 0.24649 | linear | Coding: onevsall | | | | | | | | | | Lambda: 0.0028505 | | | | | | | | | | Learner: svm | | 14 | 6 | Accept | 0.42238 | 1.5531 | 0.24649 | 0.24649 | linear | Coding: onevsall | | | | | | | | | | Lambda: 0.0016864 | | | | | | | | | | Learner: logistic | | 15 | 6 | Accept | 0.27311 | 12.485 | 0.24649 | 0.26086 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 41.969 | | | | | | | | | | Lambda: 9.9891e-06 | | 16 | 6 | Accept | 0.29584 | 2.4339 | 0.24649 | 0.26086 | linear | Coding: onevsone | | | | | | | | | | Lambda: 0.037043 | | | | | | | | | | Learner: svm | | 17 | 6 | Accept | 0.33712 | 0.26108 | 0.24649 | 0.26086 | knn | NumNeighbors: 1 | | | | | | | | | | Distance: cityblock | | 18 | 6 | Accept | 0.53365 | 2.2181 | 0.24649 | 0.26086 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.0022255 | | | | | | | | | | KernelScale: 206.47 | | 19 | 6 | Accept | 0.4834 | 1.1776 | 0.24649 | 0.26086 | knn | NumNeighbors: 72 | | | | | | | | | | Distance: correlation | | 20 | 6 | Accept | 0.46066 | 18.794 | 0.24649 | 0.26086 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 47.05 | | | | | | | | | | KernelScale: 0.46846 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 21 | 6 | Accept | 0.46066 | 19.722 | 0.24649 | 0.26086 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 47.05 | | | | | | | | | | KernelScale: 0.46846 | | 22 | 6 | Accept | 0.46066 | 19.978 | 0.24649 | 0.26086 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 47.05 | | | | | | | | | | KernelScale: 0.46846 | | 23 | 6 | Accept | 0.74185 | 0.72736 | 0.24649 | 0.26086 | tree | MinLeafSize: 1558 | | 24 | 4 | Accept | 0.25157 | 4.2545 | 0.24379 | 0.26086 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.11323 | | 25 | 4 | Best | 0.24379 | 2.7132 | 0.24379 | 0.26086 | linear | Coding: onevsone | | | | | | | | | | Lambda: 9.2928e-05 | | | | | | | | | | Learner: logistic | | 26 | 4 | Accept | 0.43255 | 0.37506 | 0.24379 | 0.26086 | discr | Delta: 0.016844 | | | | | | | | | | Gamma: 0.64466 | | 27 | 6 | Accept | 0.66796 | 0.25716 | 0.24379 | 0.26086 | knn | NumNeighbors: 77 | | | | | | | | | | Distance: jaccard | | 28 | 4 | Accept | 0.25366 | 2.6872 | 0.24379 | 0.26086 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.10033 | | 29 | 4 | Accept | 0.66796 | 0.33779 | 0.24379 | 0.26086 | knn | NumNeighbors: 77 | | | | | | | | | | Distance: jaccard | | 30 | 4 | Accept | 0.66796 | 0.29903 | 0.24379 | 0.26086 | knn | NumNeighbors: 77 | | | | | | | | | | Distance: jaccard | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 31 | 4 | Accept | 0.2773 | 0.74879 | 0.24379 | 0.26086 | tree | MinLeafSize: 94 | | 32 | 3 | Accept | 0.25606 | 11.512 | 0.24379 | 0.26086 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 132 | | | | | | | | | | LearnRate: 0.92674 | | | | | | | | | | MinLeafSize: 127 | | 33 | 3 | Accept | 0.74185 | 0.89783 | 0.24379 | 0.26086 | discr | Delta: 244.12 | | | | | | | | | | Gamma: 0.23748 | | 34 | 6 | Accept | 0.43075 | 0.10207 | 0.24379 | 0.26086 | discr | Delta: 0.0014341 | | | | | | | | | | Gamma: 0.70025 | | 35 | 6 | Accept | 0.24738 | 2.4047 | 0.24379 | 0.26086 | linear | Coding: onevsone | | | | | | | | | | Lambda: 0.0028343 | | | | | | | | | | Learner: logistic | | 36 | 6 | Accept | 0.3787 | 1.8766 | 0.24379 | 0.26086 | linear | Coding: onevsall | | | | | | | | | | Lambda: 1.4563e-06 | | | | | | | | | | Learner: logistic | | 37 | 6 | Accept | 0.53814 | 2.1153 | 0.24379 | 0.26086 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 6.8148 | | | | | | | | | | KernelScale: 382.11 | | 38 | 6 | Accept | 0.2761 | 0.22606 | 0.24379 | 0.26086 | tree | MinLeafSize: 87 | | 39 | 6 | Accept | 0.42208 | 0.77835 | 0.24379 | 0.26086 | discr | Delta: 0.0090118 | | | | | | | | | | Gamma: 0.062207 | | 40 | 6 | Accept | 0.42656 | 0.13226 | 0.24379 | 0.26086 | discr | Delta: 0.0020866 | | | | | | | | | | Gamma: 0.091054 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 41 | 6 | Accept | 0.29973 | 0.41653 | 0.24379 | 0.26086 | tree | MinLeafSize: 7 | | 42 | 6 | Accept | 0.25247 | 12.99 | 0.24379 | 0.25743 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 0.53464 | | | | | | | | | | Lambda: 2.6732e-05 | | 43 | 6 | Accept | 0.25965 | 13.385 | 0.24379 | 0.25743 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 150 | | | | | | | | | | LearnRate: 0.014842 | | | | | | | | | | MinLeafSize: 21 | | 44 | 6 | Accept | 0.28059 | 0.87454 | 0.24379 | 0.25743 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 45 | 6 | Accept | 0.27281 | 0.44508 | 0.24379 | 0.25743 | knn | NumNeighbors: 8 | | | | | | | | | | Distance: chebychev | | 46 | 6 | Accept | 0.2767 | 17.125 | 0.24379 | 0.25743 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 221 | | | | | | | | | | LearnRate: 0.0028588 | | | | | | | | | | MinLeafSize: 1 | | 47 | 5 | Accept | 0.24499 | 9.7866 | 0.24379 | 0.25743 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.019387 | | | | | | | | | | KernelScale: 0.0047515 | | 48 | 5 | Accept | 0.42178 | 0.93955 | 0.24379 | 0.25743 | discr | Delta: 6.8395e-06 | | | | | | | | | | Gamma: 0.058174 | | 49 | 6 | Best | 0.2429 | 1.5919 | 0.2429 | 0.25743 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.16719 | | | | | | | | | | KernelScale: 0.11257 | | 50 | 6 | Accept | 0.74185 | 10.422 | 0.2429 | 0.26136 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 24.681 | | | | | | | | | | Lambda: 0.092669 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 51 | 6 | Accept | 0.59049 | 3.0832 | 0.2429 | 0.26136 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 0.093586 | | | | | | | | | | Lambda: 0.0050756 | | 52 | 6 | Accept | 0.37571 | 0.17679 | 0.2429 | 0.26136 | tree | MinLeafSize: 473 | | 53 | 6 | Accept | 0.27251 | 0.16627 | 0.2429 | 0.26136 | tree | MinLeafSize: 20 | | 54 | 6 | Accept | 0.32994 | 0.26792 | 0.2429 | 0.26136 | tree | MinLeafSize: 3 | | 55 | 6 | Accept | 0.61412 | 9.4509 | 0.2429 | 0.26136 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 0.093586 | | | | | | | | | | Lambda: 0.0050756 | | 56 | 6 | Accept | 0.29973 | 0.19258 | 0.2429 | 0.26136 | tree | MinLeafSize: 7 | | 57 | 6 | Accept | 0.43225 | 0.11614 | 0.2429 | 0.26136 | discr | Delta: 0.021467 | | | | | | | | | | Gamma: 0.66016 | | 58 | 6 | Accept | 0.25725 | 29.083 | 0.2429 | 0.26136 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 304 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 100 | | 59 | 5 | Accept | 0.33144 | 1.4617 | 0.2429 | 0.26995 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.0020049 | | 60 | 5 | Accept | 0.28059 | 0.57622 | 0.2429 | 0.26995 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 61 | 6 | Accept | 0.24708 | 8.0326 | 0.2429 | 0.26995 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 4.6933 | | | | | | | | | | Lambda: 9.8945e-07 | | 62 | 6 | Accept | 0.24768 | 7.9544 | 0.2429 | 0.25978 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 4.6933 | | | | | | | | | | Lambda: 9.8945e-07 | | 63 | 6 | Accept | 0.55369 | 0.86731 | 0.2429 | 0.25978 | knn | NumNeighbors: 615 | | | | | | | | | | Distance: correlation | | 64 | 5 | Accept | 0.31947 | 13.444 | 0.2429 | 0.25978 | ensemble | Method: RUSBoost | | | | | | | | | | NumLearningCycles: 132 | | | | | | | | | | LearnRate: 0.0014516 | | | | | | | | | | MinLeafSize: 104 | | 65 | 5 | Accept | 0.57972 | 3.0436 | 0.2429 | 0.25978 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.12255 | | | | | | | | | | KernelScale: 81.172 | | 66 | 6 | Accept | 0.42596 | 0.12397 | 0.2429 | 0.25978 | discr | Delta: 9.4222e-06 | | | | | | | | | | Gamma: 0.15603 | | 67 | 6 | Accept | 0.42596 | 0.1062 | 0.2429 | 0.25978 | discr | Delta: 9.4222e-06 | | | | | | | | | | Gamma: 0.15603 | | 68 | 6 | Accept | 0.67783 | 0.82135 | 0.2429 | 0.25978 | linear | Coding: onevsall | | | | | | | | | | Lambda: 2.4732 | | | | | | | | | | Learner: logistic | | 69 | 6 | Accept | 0.25695 | 0.39058 | 0.2429 | 0.25978 | knn | NumNeighbors: 14 | | | | | | | | | | Distance: cityblock | | 70 | 6 | Accept | 0.74185 | 1.9769 | 0.2429 | 0.25978 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.94197 | | | | | | | | | | KernelScale: 524.56 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 71 | 6 | Accept | 0.6548 | 3.5066 | 0.2429 | 0.25978 | nb | DistributionNames: kernel | | | | | | | | | | Width: 2.6013 | | 72 | 6 | Accept | 0.60694 | 2.3559 | 0.2429 | 0.25764 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 0.86904 | | | | | | | | | | Lambda: 0.29724 | | 73 | 6 | Accept | 0.42926 | 0.10372 | 0.2429 | 0.25764 | discr | Delta: 0.041145 | | | | | | | | | | Gamma: 0.34864 | | 74 | 6 | Accept | 0.47383 | 0.10921 | 0.2429 | 0.25764 | tree | MinLeafSize: 793 | | 75 | 6 | Accept | 0.32456 | 0.20983 | 0.2429 | 0.25764 | tree | MinLeafSize: 5 | | 76 | 6 | Accept | 0.43703 | 1.1803 | 0.2429 | 0.25764 | linear | Coding: onevsall | | | | | | | | | | Lambda: 0.013265 | | | | | | | | | | Learner: logistic | | 77 | 6 | Accept | 0.38169 | 0.13279 | 0.2429 | 0.25764 | tree | MinLeafSize: 436 | | 78 | 6 | Accept | 0.25845 | 20.668 | 0.2429 | 0.25764 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 160 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 2 | | 79 | 6 | Accept | 0.43015 | 0.10705 | 0.2429 | 0.25764 | discr | Delta: 0.0069822 | | | | | | | | | | Gamma: 0.49526 | | 80 | 6 | Accept | 0.42208 | 0.098978 | 0.2429 | 0.25764 | discr | Delta: 0.057485 | | | | | | | | | | Gamma: 0.045714 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 81 | 6 | Accept | 0.33503 | 0.40026 | 0.2429 | 0.25764 | knn | NumNeighbors: 2 | | | | | | | | | | Distance: minkowski | | 82 | 6 | Accept | 0.52617 | 2.0596 | 0.2429 | 0.25764 | svm | Coding: onevsall | | | | | | | | | | BoxConstraint: 0.26869 | | | | | | | | | | KernelScale: 17.595 | | 83 | 6 | Accept | 0.32905 | 0.46647 | 0.2429 | 0.25764 | knn | NumNeighbors: 65 | | | | | | | | | | Distance: seuclidean | | 84 | 6 | Accept | 0.25127 | 11.058 | 0.2429 | 0.25598 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.78697 | | | | | | | | | | Lambda: 4.1197e-06 | | 85 | 6 | Accept | 0.26922 | 13.745 | 0.2429 | 0.25598 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 168 | | | | | | | | | | LearnRate: 0.0041733 | | | | | | | | | | MinLeafSize: 3 | | 86 | 6 | Accept | 0.24589 | 4.5176 | 0.2429 | 0.24982 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 1.0908 | | | | | | | | | | Lambda: 7.5756e-05 | | 87 | 6 | Accept | 0.2429 | 2.5827 | 0.2429 | 0.24982 | linear | Coding: onevsone | | | | | | | | | | Lambda: 4.7032e-08 | | | | | | | | | | Learner: svm | | 88 | 6 | Accept | 0.67873 | 4.8803 | 0.2429 | 0.24941 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 66.432 | | | | | | | | | | Lambda: 0.00097982 | | 89 | 6 | Accept | 0.28059 | 0.11127 | 0.2429 | 0.24941 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 90 | 6 | Accept | 0.25905 | 2.3728 | 0.2429 | 0.24941 | nb | DistributionNames: kernel | | | | | | | | | | Width: 0.054786 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 91 | 6 | Accept | 0.50972 | 4.7902 | 0.2429 | 0.24941 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 53 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 926 | | 92 | 6 | Accept | 0.28059 | 0.49389 | 0.2429 | 0.24941 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 93 | 6 | Accept | 0.42926 | 0.12299 | 0.2429 | 0.24941 | discr | Delta: 0.00083172 | | | | | | | | | | Gamma: 0.41309 | | 94 | 6 | Accept | 0.29704 | 1.5229 | 0.2429 | 0.24941 | ensemble | Method: AdaBoostM2 | | | | | | | | | | NumLearningCycles: 11 | | | | | | | | | | LearnRate: 0.0072731 | | | | | | | | | | MinLeafSize: 12 | | 95 | 6 | Accept | 0.28059 | 0.12303 | 0.2429 | 0.24941 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 96 | 6 | Accept | 0.28059 | 0.11495 | 0.2429 | 0.24941 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 97 | 6 | Accept | 0.29225 | 13.443 | 0.2429 | 0.24941 | ensemble | Method: RUSBoost | | | | | | | | | | NumLearningCycles: 147 | | | | | | | | | | LearnRate: 0.95321 | | | | | | | | | | MinLeafSize: 112 | | 98 | 6 | Best | 0.2411 | 2.2875 | 0.2411 | 0.24941 | linear | Coding: onevsone | | | | | | | | | | Lambda: 0.00056045 | | | | | | | | | | Learner: svm | | 99 | 6 | Accept | 0.74185 | 0.11928 | 0.2411 | 0.24941 | discr | Delta: 39.281 | | | | | | | | | | Gamma: 0.77032 | | 100 | 6 | Accept | 0.30093 | 0.13456 | 0.2411 | 0.24941 | tree | MinLeafSize: 135 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 101 | 6 | Accept | 0.46844 | 0.18889 | 0.2411 | 0.24941 | knn | NumNeighbors: 2 | | | | | | | | | | Distance: cosine | | 102 | 6 | Accept | 0.24469 | 2.2937 | 0.2411 | 0.24727 | linear | Coding: onevsone | | | | | | | | | | Lambda: 0.0012413 | | | | | | | | | | Learner: logistic | | 103 | 6 | Accept | 0.25426 | 3.6479 | 0.2411 | 0.24727 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 38 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 28 | | 104 | 6 | Accept | 0.24529 | 167.51 | 0.2411 | 0.24727 | svm | Coding: onevsone | | | | | | | | | | BoxConstraint: 0.25488 | | | | | | | | | | KernelScale: 0.0037823 | | 105 | 6 | Accept | 0.31977 | 1.9252 | 0.2411 | 0.24891 | linear | Coding: onevsone | | | | | | | | | | Lambda: 0.040102 | | | | | | | | | | Learner: logistic | | 106 | 6 | Accept | 0.50045 | 4.6994 | 0.2411 | 0.24891 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 13.293 | | | | | | | | | | Lambda: 0.0031304 | | 107 | 6 | Accept | 0.37152 | 0.14955 | 0.2411 | 0.24891 | tree | MinLeafSize: 375 | | 108 | 6 | Accept | 0.24678 | 2.9213 | 0.2411 | 0.24541 | linear | Coding: onevsone | | | | | | | | | | Lambda: 3.9918e-05 | | | | | | | | | | Learner: logistic | | 109 | 6 | Accept | 0.74484 | 4.8397 | 0.2411 | 0.24541 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.004545 | | | | | | | | | | Lambda: 0.0039369 | | 110 | 6 | Accept | 0.45947 | 0.73222 | 0.2411 | 0.24662 | linear | Coding: onevsall | | | | | | | | | | Lambda: 0.028513 | | | | | | | | | | Learner: svm | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 111 | 6 | Accept | 0.28059 | 0.11129 | 0.2411 | 0.24662 | nb | DistributionNames: normal | | | | | | | | | | Width: NaN | | 112 | 6 | Accept | 0.25845 | 7.4048 | 0.2411 | 0.24662 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 61 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 3 | | 113 | 6 | Accept | 0.42806 | 0.20953 | 0.2411 | 0.24662 | knn | NumNeighbors: 10 | | | | | | | | | | Distance: cosine | | 114 | 6 | Accept | 0.74185 | 3.5296 | 0.2411 | 0.24662 | nb | DistributionNames: kernel | | | | | | | | | | Width: 74.975 | | 115 | 6 | Accept | 0.29106 | 21.854 | 0.2411 | 0.24662 | ensemble | Method: Bag | | | | | | | | | | NumLearningCycles: 249 | | | | | | | | | | LearnRate: NaN | | | | | | | | | | MinLeafSize: 292 | | 116 | 6 | Accept | 0.77864 | 10.34 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.0050713 | | | | | | | | | | Lambda: 3.7406e-06 | | 117 | 6 | Accept | 0.24708 | 5.2433 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.8979 | | | | | | | | | | Lambda: 0.00028514 | | 118 | 6 | Accept | 0.71762 | 4.9096 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 500.74 | | | | | | | | | | Lambda: 7.9516e-05 | | 119 | 6 | Accept | 0.7601 | 12.416 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.013078 | | | | | | | | | | Lambda: 0.00023088 | | 120 | 6 | Accept | 0.31289 | 3.9454 | 0.2411 | 0.24662 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 4.8535 | | | | | | | | | | Lambda: 0.00015045 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 121 | 6 | Accept | 0.68771 | 3.8043 | 0.2411 | 0.24662 | kernel | Coding: onevsall | | | | | | | | | | KernelScale: 553.29 | | | | | | | | | | Lambda: 2.0294e-05 | | 122 | 6 | Accept | 0.24948 | 6.555 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.69606 | | | | | | | | | | Lambda: 5.6834e-05 | | 123 | 6 | Accept | 0.24649 | 6.265 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 1.5014 | | | | | | | | | | Lambda: 7.4113e-05 | | 124 | 6 | Accept | 0.24649 | 6.2611 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.75029 | | | | | | | | | | Lambda: 8.0356e-05 | | 125 | 6 | Accept | 0.25217 | 10.816 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.8744 | | | | | | | | | | Lambda: 3.9465e-06 | | 126 | 6 | Accept | 0.25755 | 0.25578 | 0.2411 | 0.24662 | knn | NumNeighbors: 133 | | | | | | | | | | Distance: minkowski | | 127 | 6 | Accept | 0.24589 | 5.6905 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 1.0841 | | | | | | | | | | Lambda: 0.00019 | | 128 | 6 | Accept | 0.25366 | 6.0616 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 13.915 | | | | | | | | | | Lambda: 4.7818e-05 | | 129 | 5 | Accept | 0.3111 | 19.355 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.23685 | | | | | | | | | | Lambda: 3.4559e-07 | | 130 | 5 | Accept | 0.24649 | 6.139 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.89435 | | | | | | | | | | Lambda: 0.00018696 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 131 | 6 | Accept | 0.24798 | 5.2185 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 1.9045 | | | | | | | | | | Lambda: 0.00021312 | | 132 | 6 | Accept | 0.24828 | 5.4701 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 1.3796 | | | | | | | | | | Lambda: 0.00034835 | | 133 | 6 | Accept | 0.28896 | 8.0051 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.18555 | | | | | | | | | | Lambda: 5.7486e-05 | | 134 | 6 | Accept | 0.24738 | 0.22382 | 0.2411 | 0.24662 | knn | NumNeighbors: 98 | | | | | | | | | | Distance: minkowski | | 135 | 6 | Accept | 0.24678 | 6.1613 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 3.8708 | | | | | | | | | | Lambda: 0.00022901 | | 136 | 5 | Accept | 0.24529 | 5.7754 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.98622 | | | | | | | | | | Lambda: 0.00022785 | | 137 | 5 | Accept | 0.277 | 0.63033 | 0.2411 | 0.24662 | knn | NumNeighbors: 197 | | | | | | | | | | Distance: minkowski | | 138 | 6 | Accept | 0.25636 | 5.2351 | 0.2411 | 0.24662 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 5.9871 | | | | | | | | | | Lambda: 0.00025502 | | 139 | 6 | Accept | 0.24589 | 5.4419 | 0.2411 | 0.24661 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.71386 | | | | | | | | | | Lambda: 0.00023876 | | 140 | 6 | Accept | 0.24828 | 5.1049 | 0.2411 | 0.24617 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 0.93977 | | | | | | | | | | Lambda: 0.00056651 | |===========================================================================================================================================| | Iter | Active | Eval | Validation | Time for training | Observed min | Estimated min | Learner | Hyperparameter: Value | | | workers | result | loss | & validation (sec)| validation loss | validation loss | | | |===========================================================================================================================================| | 141 | 6 | Accept | 0.27879 | 5.6359 | 0.2411 | 0.24618 | kernel | Coding: onevsone | | | | | | | | | | KernelScale: 21.38 | | ...
__________________________________________________________ Optimization completed. Total iterations: 240 Total elapsed time: 608.1602 seconds Total time for training and validation: 1585.7079 seconds Best observed learner is a multiclass svm model with: Coding (ECOC): onevsone BoxConstraint: 0.85149 KernelScale: 0.48514 Observed validation loss: 0.23961 Time for training and validation: 1.2946 seconds Best estimated learner (returned model) is a multiclass svm model with: Coding (ECOC): onevsone BoxConstraint: 912.11 KernelScale: 13.622 Estimated validation loss: 0.24168 Estimated time for training and validation: 1.1953 seconds Documentation for fitcauto display
Итоговая модель возвращена fitcauto соответствует лучшему предполагаемому ученику. Прежде, чем возвратить модель, функция переобучает его с помощью целых обучающих данных (creditTrain), перечисленный Learner (или модель) тип и отображенные значения гиперпараметра.
Оцените эффективность набора тестов
Модель Mdl соответствует лучшей точке в Байесовой оптимизации согласно "min-visited-mean" критерий. Чтобы измерить, как модель выполнит на новых данных, посмотрите на наблюдаемую точность перекрестной проверки модели (cvAccuracy) и его общая предполагаемая эффективность на основе Байесовой оптимизации (estimatedAccuracy).
[x,~,iteration] = bestPoint(Results,"Criterion","min-visited-mean"); cvError = Results.ObjectiveTrace(iteration); cvAccuracy = 1 - cvError
cvAccuracy = 0.7598
estimatedError = predictObjective(Results,x); estimatedAccuracy = 1 - estimatedError
estimatedAccuracy = 0.7583
Оцените эффективность модели на наборе тестов. Создайте матрицу беспорядка из результатов и задайте порядок классов в матрице беспорядка.
testAccuracy = 1 - loss(Mdl,creditTest,"Rating")testAccuracy = 0.7438
cm = confusionchart(creditTest.Rating,predict(Mdl,creditTest)); sortClasses(cm,["AAA","AA","A","BBB","BB","B","CCC"])
В зависимости от размера вашего набора данных, количества учеников, которых вы задаете, и метод оптимизации, который вы выбираете, fitcauto может занять время, чтобы запуститься.
Если у вас есть лицензия Parallel Computing Toolbox, можно ускорить расчеты путем выполнения оптимизации параллельно. Для этого задайте "HyperparameterOptimizationOptions",struct("UseParallel",true). Можно включать дополнительные поля в структуру, чтобы управлять другими аспектами оптимизации. Смотрите HyperparameterOptimizationOptions.
Если fitcauto с Байесовой оптимизацией занимает много времени, чтобы запуститься из-за количества наблюдений в вашем наборе обучающих данных (например, более чем 10 000), рассмотреть использование fitcauto с оптимизацией ASHA вместо этого. Оптимизация ASHA часто находит хорошие решения быстрее, чем Байесова оптимизация для наборов данных со многими наблюдениями. Чтобы использовать оптимизацию ASHA, задайте "HyperparameterOptimizationOptions",struct("Optimizer","asha"). Можно включать дополнительные поля в структуру, чтобы управлять другими аспектами оптимизации. В частности, если вы имеете ограничение времени, задаете MaxTime поле HyperparameterOptimizationOptions структура, чтобы ограничить номер секунд fitcauto запуски.
Если вы не уверены, какие модели работают лучше всего на ваш набор данных, можно альтернативно использовать приложение Classification Learner. Используя приложение, можно выполнить гиперпараметр, настраивающийся для различных моделей, и выбрать оптимизированную модель, которая выполняет лучше всего. Несмотря на то, что необходимо выбрать определенную модель, прежде чем можно будет настроить гиперпараметры модели, Classification Learner обеспечивает большую гибкость для выбора optimizable гиперпараметров и устанавливания значений гиперпараметра. Однако вы не можете оптимизировать параллельно, оптимизировать "linear" или "kernel" ученики, задайте веса наблюдения, задайте априорные вероятности или используйте оптимизацию ASHA в приложении. Для получения дополнительной информации смотрите Гипероптимизацию параметров управления в Приложении Classification Learner.
Если вы знаете, какие модели могут удовлетворить вашим данным, можно альтернативно использовать соответствующие подходящие функции модели и задать OptimizeHyperparameters аргумент значения имени, чтобы настроить гиперпараметры. Можно сравнить результаты через модели, чтобы выбрать лучший классификатор. Для примера этого процесса смотрите, что Двигающий Выбор Модели Автоматизации Использует Байесовую Оптимизацию.
[1] Литий, Лиам, Кевин Джемисон, Afshin Rostamizadeh, Екатерина Гонина, Мориц Хардт, Бенджамин Речт и Амит Тэлволкэр. “Система для Настройки Гиперпараметра С массовым параллелизмом”. ArXiv:1810.05934v5 [Cs], 16 марта 2020. https://arxiv.org/abs/1810.05934v5.
fitcdiscr | fitcecoc | fitcensemble | fitcknn | fitclinear | fitcnb | fitcsvm | fitctree | fitckernel