ISlSerie.BinningSettings

Syntax

Description

The BinningSettings property returns parameters of the Binning procedure.

Comments

Binning is a procedure of data preparation for data mining methods working with categorical data.

The method idea is that input array of numerical data is split into specified number of ranges according to the specified splitting rules. Obtained ranges are further used in data mining analysis as single categories.

To get results of the Binning procedure execution, use the ISlSerie.Categories property.

Example

Add links to the system assemblies: MathFin, Stat.

Sub UserProc;
Var
    method: SmDecisionTree;
    y, x1, x2, x3, x4: array[15] Of double;
    Explanatories: ISlSeries;
    Dependent, FilledDependent: ISlSerie;
    BinningSettings: IBinningSettings;
    res: integer;
    maxlength, i: integer;
    strng: string;
    leng: array[4] Of integer;
    Function AddStrVal(str, strval: string): string;
    Begin
        Return str + "  " + strval;
    End Function AddStrVal;
Begin
    // Create object to calculate method
    method := New SmDecisionTree.Create;
    // Set variables initial values
    x1[0] := 2; x2[0] := 2; x3[0] := 4; x4[0] := 1; y[0] := 1;
    x1[1] := 5; x2[1] := 4; x3[1] := 5; x4[1] := 2; y[1] := 2;
    x1[2] := 15; x2[2] := 6; x3[2] := 6; x4[2] := 3; y[2] := 3;
    x1[3] := 3; x2[3] := 1; x3[3] := 34; x4[3] := 3; y[3] := 1;
    x1[4] := 8; x2[4] := 3; x3[4] := 7; x4[4] := 2; y[4] := 2;
    x1[5] := 11; x2[5] := 5; x3[5] := 5; x4[5] := 1; y[5] := 3;
    x1[6] := 1; x2[6] := 2; x3[6] := 3; x4[6] := 3; y[6] := 1;
    x1[7] := 6; x2[7] := 4; x3[7] := 5; x4[7] := 2; y[7] := 2;
    x1[8] := 12; x2[8] := 6; x3[8] := 12; x4[8] := 3; y[8] := 3;
    x1[9] := 4; x2[9] := 2; x3[9] := 8; x4[9] := 2; y[9] := 1;
    x1[10] := 7; x2[10] := 4; x3[10] := 13; x4[10] := 2; y[10] := 2;
    x1[11] := 13; x2[11] := 6; x3[11] := 6; x4[11] := 2; y[11] := 3;
    x1[12] := 1; x2[12] := 1; x3[12] := 9; x4[12] := 1; y[12] := Double.Nan;
    x1[13] := 9; x2[13] := 4; x3[13] := 6; x4[13] := 1; y[13] := Double.Nan;
    x1[14] := 11; x2[14] := 7; x3[14] := 5; x4[14] := 1; y[14] := Double.Nan;
    // Set explained series
    method.Dependent.Value := y;
    // Define explanatory series
    Explanatories := method.Explanatories;
    Explanatories.Add.Value := x1;
    Explanatories.Add.Value := x2;
    Explanatories.Add.Value := x3;
    Explanatories.Add.Value := x4;
    // Get parameters of the Binning procedure of the explained series
    Dependent := method.Dependent;
    BinningSettings := Dependent.BinningSettings;
    // Set binning method
    BinningSettings.Method := BinningMethod.EqualDepth;
    // Set number of resulting categories
    BinningSettings.NumOfCategories := 4;
    // Set maximum number of iterations
    BinningSettings.MaxIt := 9;
    // Execute method calculation
    res := method.Execute;
    // Display classification results
    Debug.WriteLine(" == Classification == ");
    Debug.WriteLine(" No    in  in_cat  out_cat out");
    FilledDependent := method.FilledDependent;
    leng[0] := Dependent.Value.Length;
    leng[1] := Dependent.Categories.Length;
    leng[2] := FilledDependent.Categories.Length;
    leng[3] := FilledDependent.Value.Length;
    maxlength := Math.MaxI(leng);
    For i := 0 To maxlength - 1 Do
        strng := i.ToString;
        If (i < leng[0]) Then strng := addstrval(strng, Dependent.Value[i].ToString);
        Else strng := addstrval(strng, "-"); End If;
        If (i < leng[1]) Then strng := addstrval(strng, Dependent.Categories[i].ToString);
        Else strng := addstrval(strng, "-"); End If;
        If (i < leng[2]) Then strng := addstrval(strng, FilledDependent.Categories[i].ToString);
        Else strng := addstrval(strng, "-"); End If;
        If (i < leng[3]) Then strng := addstrval(strng, FilledDependent.Value[i].ToString);
        Else strng := addstrval(strng, "-"); End If;
        Debug.WriteLine(strng);
    End For;
End Sub UserProc;

After executing the example substitution of missing data will be executed in series values using decision tree. To split explanatory series into categories, the Binning procedure will be used.