7eaddd9a54
ML - устранение неточностей и багов ML - оптимизация производительности DecisionTreeRegressor.Fit, RandomForestRegressor.Fit ML - тесты
49 lines
1.9 KiB
ObjectPascal
49 lines
1.9 KiB
ObjectPascal
uses MLABC;
|
|
|
|
begin
|
|
var ds := Datasets.TitanicRu;
|
|
var df := ds.Data.Drop(['Id', 'Имя']);
|
|
|
|
// Заполняем пропуски.
|
|
var ageImputer := new Imputer(['Возраст']);
|
|
df := ageImputer.FitTransform(df);
|
|
|
|
var portImputer := new Imputer('Саутгемптон', ['ПортПосадки']);
|
|
df := portImputer.FitTransform(df);
|
|
|
|
// Кодируем категориальные признаки числами.
|
|
var sexEncoder := new LabelEncoder('Пол');
|
|
df := sexEncoder.FitTransform(df);
|
|
|
|
var portEncoder := new LabelEncoder('ПортПосадки');
|
|
df := portEncoder.FitTransform(df);
|
|
|
|
var features := ['Класс', 'Пол', 'Возраст', 'БратьяИСупруги', 'РодителиИДети', 'ЦенаБилета', 'ПортПосадки'];
|
|
var X := df.ToMatrix(features);
|
|
var y := df.GetIntColumn('Выжил');
|
|
|
|
var (Xtrain, Xtest, ytrain, ytest) := Validation.TrainTestSplit(X, y, testRatio := 0.2, seed := 42);
|
|
|
|
var scaler := new StandardScaler;
|
|
scaler.Fit(Xtrain);
|
|
var XtrainScaled := scaler.Transform(Xtrain);
|
|
var XtestScaled := scaler.Transform(Xtest);
|
|
|
|
var lr := new LogisticRegression(learningRate := 0.01, epochs := 2000);
|
|
lr.Fit(XtrainScaled, ytrain);
|
|
var predLR := lr.Predict(XtestScaled);
|
|
|
|
var tree := new DecisionTreeClassifier(maxDepth := 5, minSamplesLeaf := 3, minSamplesSplit := 6);
|
|
tree.Fit(Xtrain, ytrain);
|
|
var predTree := tree.Predict(Xtest);
|
|
|
|
var forest := new RandomForestClassifier(nTrees := 100, maxDepth := 6, minSamplesLeaf := 3, minSamplesSplit := 6);
|
|
forest.Fit(Xtrain, ytrain);
|
|
var predForest := forest.Predict(Xtest);
|
|
|
|
Println('Сравнение моделей на TitanicRu');
|
|
Println($'LogisticRegression: Accuracy = {Metrics.Accuracy(ytest, predLR):F3}');
|
|
Println($'DecisionTreeClassifier: Accuracy = {Metrics.Accuracy(ytest, predTree):F3}');
|
|
Println($'RandomForestClassifier: Accuracy = {Metrics.Accuracy(ytest, predForest):F3}');
|
|
end.
|