33 lines
1.1 KiB
ObjectPascal
33 lines
1.1 KiB
ObjectPascal
|
// В этом примере сравниваются две модели на нелинейной задаче классификации:
|
|||
|
|
// LogisticRegression и DecisionTreeClassifier.
|
|||
|
|
|
|||
|
|
uses MLABC;
|
|||
|
|
|
|||
|
|
begin
|
|||
|
|
var (X, y) := Datasets.MakeMoons(
|
|||
|
|
n := 450,
|
|||
|
|
noise := 0.20,
|
|||
|
|
seed := 42
|
|||
|
|
);
|
|||
|
|
|
|||
|
|
var (XTrain, XTest, yTrain, yTest) := Validation.TrainTestSplit(X, y, 0.25, seed := 42);
|
|||
|
|
|
|||
|
|
var logreg := new LogisticRegression(learningRate := 0.05, epochs := 1000);
|
|||
|
|
logreg.Fit(XTrain, yTrain);
|
|||
|
|
var logregAcc := ClassificationMetrics.Accuracy(yTest, logreg.Predict(XTest));
|
|||
|
|
|
|||
|
|
var tree := new DecisionTreeClassifier(
|
|||
|
|
maxDepth := 5,
|
|||
|
|
minSamplesSplit := 6,
|
|||
|
|
minSamplesLeaf := 3,
|
|||
|
|
seed := 42
|
|||
|
|
);
|
|||
|
|
tree.Fit(XTrain, yTrain);
|
|||
|
|
var treeAcc := ClassificationMetrics.Accuracy(yTest, tree.Predict(XTest));
|
|||
|
|
|
|||
|
|
Println('Сравнение моделей на нелинейной задаче классификации');
|
|||
|
|
Println;
|
|||
|
|
Println($'Логистическая регрессия: Accuracy = {logregAcc:F3}');
|
|||
|
|
Println($'Дерево решений: Accuracy = {treeAcc:F3}');
|
|||
|
|
end.
|