33 lines
901 B
ObjectPascal
33 lines
901 B
ObjectPascal
|
// В этом примере MatrixPipeline используется
|
|||
|
|
// для задачи регрессии.
|
|||
|
|
|
|||
|
|
uses MLABC;
|
|||
|
|
|
|||
|
|
begin
|
|||
|
|
var ds := Datasets.MoscowHousing;
|
|||
|
|
|
|||
|
|
var features := ['area', 'rooms', 'floor', 'max_floor', 'distance_to_metro', 'building_age'];
|
|||
|
|
|
|||
|
|
var (trainDs, testDs) := ds.TrainTestSplit(testRatio := 0.2, seed := 42);
|
|||
|
|
|
|||
|
|
var XTrain := trainDs.Data.ToMatrix(features);
|
|||
|
|
var yTrain := trainDs.Data.ToVector(ds.Target);
|
|||
|
|
|
|||
|
|
var XTest := testDs.Data.ToMatrix(features);
|
|||
|
|
var yTest := testDs.Data.ToVector(ds.Target);
|
|||
|
|
|
|||
|
|
var pipe :=
|
|||
|
|
MatrixPipeline.Build(
|
|||
|
|
new StandardScaler,
|
|||
|
|
new LinearRegression
|
|||
|
|
);
|
|||
|
|
|
|||
|
|
pipe.Fit(XTrain, yTrain);
|
|||
|
|
var pred := pipe.Predict(XTest);
|
|||
|
|
var r2 := RegressionMetrics.R2(yTest, pred);
|
|||
|
|
|
|||
|
|
Println('MatrixPipeline для задачи регрессии');
|
|||
|
|
Println;
|
|||
|
|
Println($'R² на тестовой выборке: {r2:F3}');
|
|||
|
|
end.
|