TMtxLogistReg Class — Stats Master

type TMtxLogistReg = class(TMtxComponent);

Performs logistic regression.

Use TMtxLogistReg component to perform ordinary or ordinal logistic regression on dataset.Suppose y takes values in k ordered categories, and let p_ij be the cumulative probability that y(i) falls in the j'th category or higher. The ordinal logistic regression model is defined as:

logit(p_ij) = theta(j) + A_i'B , i = 1,..,length(Y), j = 1,..,k-1,

where A_i is the i'th row of A . The number of ordinal categories k is taken to be the number of distinct values of int(y). If k is 2 (two categories) the model is ordinary logistic regression.

How to use TMtxLogistReg component?

Drop a TMtxLogistReg component on the form.
Define TMtxLogistReg.A, the matrix of independent variables.
Define TMtxLogistReg.Y, grouping (categories) vector. Note: All values have to be integers.
Define initial estimates for TMtxLogistReg.Theta and TMtxLogistReg.B coefficients. Alternatively you can also set TMtxLogistReg.AutoInitEstimates to true and let the algorithm calculate initial estimates for Theta and B.
(Optionally) Set TMtxLogistReg.Tolerance to define desired B, Theta estimates precision. Set TMtxLogistReg.MaxIteration to define maximum number of iterations in main calculation loop.
Call the TMtxLogistReg.Recalc method to trigger calculation.

Results:

TMtxLogistReg.B : B coefficients estimates.
TMtxLogistReg.Theta : Theta coefficients estimates.
TMtxLogistReg.TBStdErr : Theta and B coefficients estimates standard errors.

Examples

delphi

Uses ..., StatTools, MtxVecEdit;
var TestComp: TMtxLogistReg;
begin
    TestComp := TMtxLogistReg.Create(Self);
    try
        TestComp.A.SetIt(27,3,false,
        [0.8, 1.9, 0.996,
        0.9, 1.4, 0.992,
        0.8, 0.8, 0.982,
        1, 0.7, 0.986,
        0.9,    1.3, 0.98,
        1,      0.6, 0.982,
        0.95,   1,      0.992,
        0.95,   1.9,    1.02,
        1,      0.8,    0.999,
        0.95,   0.5, 1.038,
        0.85, 0.7, 0.988,
        0.7,    1.2, 0.982,
        0.8,    0.4, 1.006,
        0.2,    0.8, 0.99,
        1,      1.1, 0.99,
        1,      1.9, 1.02,
        0.65, 0.5,      1.014,
        1,      1,      1.004,
        0.5,    0.6, 0.99,
        1,      1.1, 0.986,
        1,      0.4,    1.01,
        0.9,    0.6, 1.02,
        1,      1,      1.002,
        0.95, 1.6, 0.988,
        1,      1.7, 0.99,
        1,      0.9, 0.986,
        1,      0.7, 0.986]);
        TestComp.Y.SetIt(false,[1,1,0,0,1,0,1,0,0,0,0,0,0,0,0,1,0,0,0,1,0,0,1,0,1,1,0]);
        TestComp.Recalc;
        // Results =>
        // B = (9.65215222458842, 3.86710032907408, -82.073774279211)
        // Theta = (-67.6339061278272)
        // TBStdErr = (56.8875416435276, 7.75107606604495, 1.77827769017976, 61.712376172072)
        // meaning Theta StdErr = 56.8875416435276,
        //         Beta StdErrs = 7.75107606604495, 1.77827769017976, 61.712376172072
        ViewValues(TestComp.b,'B',true);
        ViewValues(TestComp.Theta,'Theta',true);
        ViewValues(TestComp.TBStdErr,'SE',true);
    finally
        TestComp.Free;
    end;
end;

Properties

Name	Type	Description
A	TMtx	Independent variables.
Alpha	Double	Desired significant value for the statistical tests.
AutoInitEstimates	Boolean	Automatically calculate initial estimates for regression coefficients.
AutoUpdate	Boolean	If true then changing any of the TMtxLogistReg properties will trigger the TMtxLogistReg.Recalc method.
B	TVec	B coefficients estimates.
BlockAssign	Boolean	Block streaming of specific properties when storing only a "template".
Dirty	Boolean	When any of the TMtxLogistReg properties changes, this property is automatically set to true.
EditorActive	Boolean	Returns True, if the component editor is displayed.
MaxIteration	Integer	Defines one of the conditions for terminating the regression coefficient calculation.
Reference	TReferenceList	Stores a list of components that have to be notified, when this component is destroyed.
TBStdErr	TVec	Standard errors of Theta, B estimates.
Theta	TVec	Theta coefficients estimates.
Tolerance	Double	Defines one of the conditions for terminating the regression coefficient calculation.
Y	TVec	Grouping variable.

Methods

Name	Description
AllowStreaming	Returns the negatated value of BlockAssign.
Assign	Assign values of all published properties from Source.
AssignTemplate	Assign values of "template-like" published properties from Source.
EditorClass	To be overriden in descendanr classses.
LoadFromStream	Load the component from Src stream.
LoadTemplateFromFile	Load a template from file.
LoadTemplateFromStream	Load a template from stream.
Recalc	Triggers logistic regression recalculation.
Reset	Reset is called after loading the data from stream or file.
SaveTemplateToFile	Save template to file.
SaveTemplateToStream	Save template to stream.
SaveToStream	Save the component to Dst stream.