LazStats: Refactor PlotXYUnit to use Calc_BivariateRegression()

git-svn-id: https://svn.code.sf.net/p/lazarus-ccr/svn@7758 8e941d3f-bd1b-0410-a28a-d453659cc2b4

applications/lazstats/source/LazStats.lpi

@@ -264,7 +264,7 @@
       <Unit20>
         <Filename Value="forms\analysis\descriptive\plotxyunit.pas"/>
         <IsPartOfProject Value="True"/>
-        <ComponentName Value="PlotXYFrm"/>
+        <ComponentName Value="PlotXYForm"/>
         <HasResources Value="True"/>
         <ResourceBaseClass Value="Form"/>
         <UnitName Value="PlotXYUnit"/>

applications/lazstats/source/forms/analysis/descriptive/plotxyunit.lfm

@@ -1,4 +1,4 @@
-inherited PlotXYFrm: TPlotXYFrm
+inherited PlotXYForm: TPlotXYForm
   Left = 427
   Height = 500
   Top = 175

applications/lazstats/source/forms/analysis/descriptive/plotxyunit.pas

@@ -8,15 +8,15 @@ interface
 
 uses
   Classes, SysUtils, FileUtil, Forms, Controls, Graphics, Dialogs,
-  StdCtrls, ExtCtrls, Buttons, ComCtrls,
+  StdCtrls, ExtCtrls, Buttons, ComCtrls, Grids,
-  MainUnit, Globals, FunctionsLib, BasicStatsReportAndChartFormUnit,
+  MainUnit, Globals, MathUnit, BasicStatsReportAndChartFormUnit,
   ReportFrameUnit, ChartFrameUnit;
 
 type
 
-  { TPlotXYFrm }
+  { TPlotXYForm }
 
-  TPlotXYFrm = class(TBasicStatsReportAndChartForm)
+  TPlotXYForm = class(TBasicStatsReportAndChartForm)
     ConfEdit: TEdit;
     Label4: TLabel;
     LineChk: TCheckBox;
@@ -41,15 +41,20 @@ type
     procedure YOutBtnClick(Sender: TObject);
 
   private
-    procedure PlotXY(XPoints, YPoints, UpConf, LowConf: DblDyneVec;
+    procedure PlotXY(XPoints, YPoints: DblDyneVec;
-      XMean, YMean, R, Slope, Intercept: Double);
+      const ARegressionResults: TBivariateRegressionResults);
+
+    function PrepareData(ADataGrid: TStringGrid;
+      out xCol, ycol: Integer; out XData, YData: DblDyneVec;
+      out ColNoSelected: IntDyneVec): Boolean;
+
+    procedure WriteToReport(ARegressionResults: TBivariateRegressionResults);
 
   protected
     procedure AdjustConstraints; override;
     procedure Compute; override;
     procedure UpdateBtnStates; override;
-    function Validate(out AMsg: String; out AControl: TWinControl;
+    function Validate(out AMsg: String; out AControl: TWinControl): Boolean; override;
-      Xcol,Ycol: Integer): Boolean; reintroduce;
 
   public
     constructor Create(AOwner: TComponent); override;
@@ -57,7 +62,8 @@ type
   end; 
 
 var
-  PlotXYFrm: TPlotXYFrm;
+  PlotXYForm: TPlotXYForm;
+
 
 implementation
 
@@ -65,12 +71,12 @@ implementation
 
 uses
   TAChartUtils, TAChartAxisUtils, TALegend, TASources, TACustomSeries, TASeries,
-  MathUnit, GridProcs, Utils;
+  GridProcs, Utils;
 
 
-{ TPlotXYFrm }
+{ TPlotXYForm }
 
-constructor TPlotXYFrm.Create(AOwner: TComponent);
+constructor TPlotXYForm.Create(AOwner: TComponent);
 begin
   inherited;
 
@@ -98,7 +104,7 @@ begin
 end;
 
 
-procedure TPlotXYFrm.AdjustConstraints;
+procedure TPlotXYForm.AdjustConstraints;
 begin
   ParamsPanel.Constraints.MinHeight := OptionsGroup.Top + OptionsGroup.Height +
     OptionsGroup.BorderSpacing.Bottom + ButtonBevel.Height +
@@ -110,26 +116,19 @@ begin
 end;
 
 
-procedure TPlotXYFrm.Compute;
+procedure TPlotXYForm.Compute;
 var
-  xMean, yMean, xVariance, yVariance, xStddev, yStddev: double;
-  SXX, SXY, SYY, R, slope, intercept, t, confBand: Double;
-  sePred, predicted, sedata: double;
-  i, xCol, yCol, N, DF: integer;
   xValues: DblDyneVec = nil;
   yValues: DblDyneVec = nil;
-  UpConf: DblDyneVec = nil;
-  lowConf: DblDyneVec = nil;
   ColNoSelected: IntDyneVec= nil;
+  xCol, yCol: Integer;
+  confBand: Double;
+  regressionRes: TBivariateRegressionResults;
   C: TWinControl;
   msg: String;
-  lReport: TStrings;
 begin
-  xCol := OS3MainFrm.DataGrid.Rows[0].IndexOf(XEdit.Text);
+  // Validation: Make sure that XEdit and YEdit are not empty
-  yCol := OS3MainFrm.DataGrid.Rows[0].IndexOf(YEdit.Text);
+  if not Validate(msg, C) then
-
-  // Validation
-  if not Validate(msg, C, Xcol, Ycol) then
   begin
     C.SetFocus;
     ErrorMsg(msg);
@@ -137,99 +136,36 @@ begin
     exit;
   end;
 
-  SetLength(ColNoSelected, 2);
+  // Extract data from Grid
-  ColNoSelected[0] := Xcol;
+  if not PrepareData(OS3MainFrm.DataGrid, xCol, yCol, xValues, yValues, ColNoSelected) then
-  ColNoSelected[1] := Ycol;
-
-  xValues := CollectValues(OS3MainFrm.DataGrid, xCol, ColNoSelected);
-  yValues := CollectValues(OS3MainFrm.DataGrid, yCol, ColNoSelected);
-  if (Length(yValues) = 0) then
-  begin
-    ErrorMsg('No y data');
     exit;
-  end;
-  if (Length(xValues) <> Length(yValues)) then
-  begin
-    ErrorMsg('Different count of x and y values.');
-    exit;
-  end;
-  N := Length(yValues);
 
-  Calc_MeanVarStddevSS(xValues, xMean, xVariance, xStdDev, SXX);
+  // Sort on x values
-  Calc_MeanVarStddevSS(yValues, yMean, yVariance, yStdDev, SYY);
+  SortOnX(xValues, yValues);
-  SXY := 0;
-  for i := 0 to N-1 do
-    SXY := SXY + xValues[i] * yValues[i];
 
-  R := (SXY - xMean * yMean * N) / ((N - 1) * xStdDev * yStdDev);
+  // Calculate regression
-  sePred := sqrt(1.0 - sqr(R)) * yStdDev * sqrt((N - 1) / (N - 2));
+  confBand := StrToFloat(ConfEdit.Text) / 100.0;
-  slope := R * yStdDev / xStdDev;
+  Calc_BivariateRegression(xValues, yValues, confBand, regressionRes);
-  intercept := yMean - slope * xMean;
 
   // Print the descriptive statistics to the output frame
-  lReport := TStringList.Create;
+  WriteToReport(regressionRes);
-  try
-    lReport.Add('X vs. Y PLOT');
-    lReport.Add('');
-    lReport.Add('Data file: %s', [OS3MainFrm.FileNameEdit.Text]);
-    lReport.Add('');
-    lReport.Add('Variables:');
-    lReport.Add('  X: %s', [XEdit.Text]);
-    lReport.Add('  Y: %s', [YEdit.Text]);
-    lReport.Add('');
-    lReport.Add('Variable      Mean    Variance  Std.Dev.');
-    lReport.Add('----------  --------  --------  --------');
-    lReport.Add('%-10s  %8.2f  %8.2f  %8.2f', [XEdit.Text, XMean, XVariance, XStdDev]);
-    lReport.Add('%-10s  %8.2f  %8.2f  %8.2f', [YEdit.Text, YMean, YVariance, YStdDev]);
-    lReport.Add('');
-    lReport.Add('Regression:');
-    lReport.Add('  Correlation:                %8.3f', [R]);
-    lReport.Add('  Slope:                      %8.3f', [slope]);
-    lReport.Add('  Intercept:                  %8.3f', [intercept]);
-    lReport.Add('  Standard Error of Estimate: %8.3f', [sePred]);
-    lReport.Add('  Number of good cases:       %8d',   [N]);
-
-    FReportFrame.DisplayReport(lReport);
-  finally
-    lReport.Free;
-  end;
-
-  // Get upper and lower confidence points for each X value
-  if ConfChk.Checked then
-  begin
-    SortOnX(xValues, yValues);
-
-    SetLength(UpConf, N);
-    SetLength(lowConf, N);
-
-    confBand := StrToFloat(ConfEdit.Text) / 100.0;
-    DF := N - 2;
-    t := InverseT(confBand, DF);
-    for i := 0 to N-1 do
-    begin
-      predicted := intercept + slope * xValues[i];
-      seData := sePred * sqrt(1.0 + 1/N + sqr(xValues[i] - XMean) / SXX);
-      upConf[i]  := predicted + t * seData;
-      lowConf[i] := predicted - t * seData;
-    end;
-  end
-  else
-    confBand := 0.0;
 
   // Plot the values (and optional line and confidence band if elected)
-  PlotXY(xValues, yValues, upConf, lowConf, xMean, yMean, R, slope, intercept);
+  PlotXY(xValues, yValues, regressionRes);
 end;
 
 
-procedure TPlotXYFrm.PlotXY(XPoints, YPoints, UpConf, LowConf: DblDyneVec;
+procedure TPlotXYForm.PlotXY(XPoints, YPoints: DblDyneVec;
-  XMean, YMean, R, Slope, Intercept: Double);
+  const ARegressionResults: TBivariateRegressionResults);
 var
-  tmpX: array of Double = nil;
+  tmpX: DblDyneVec = nil;
-  tmpY: array of Double = nil;
+  tmpY: DblDyneVec = nil;
+  conf: DblDyneVec = nil;
   xmin, xmax, ymin, ymax: Double;
   rightLabels: TListChartSource;
   topLabels: TListChartSource;
   ser: TChartSeries;
+  i: Integer;
 begin
   rightLabels := FChartFrame.Chart.AxisList[2].Marks.Source as TListChartSource;
   rightLabels.Clear;
@@ -239,16 +175,20 @@ begin
 
   // Titles
   FChartFrame.SetTitle('X vs. Y plot using file ' + OS3MainFrm.FileNameEdit.Text);
-  FChartFrame.SetFooter(Format('R(X,Y) = %.3f, Slope = %.3f, Intercept = %.3f', [
+  with ARegressionResults do
-    R, Slope, Intercept
+    FChartFrame.SetFooter(Format('R(X,Y) = %.3f, Slope = %.3f, Intercept = %.3f', [
-  ]));
+      R, Slope, Intercept
+    ]));
   FChartFrame.SetXTitle(XEdit.Text);
   FChartFrame.SetYTitle(YEdit.Text);
 
   // Draw upper confidence band
   if ConfChk.Checked then
   begin
-    ser := FChartFrame.PlotXY(ptLines, XPoints, UpConf, nil, nil, 'Upper confidence band', clRed);
+    SetLength(conf, ARegressionResults.Count);
+     for i := 0 to High(conf) do
+       conf[i] := ARegressionResults.ConfidenceLimits(XPoints[i], true);
+    ser := FChartFrame.PlotXY(ptLines, XPoints, conf, nil, nil, 'Upper confidence band', clRed);
     rightLabels.Add(ser.yValue[ser.Count-1], ser.YValue[ser.Count-1], 'UCL');
   end;
 
@@ -258,7 +198,10 @@ begin
   // Draw lower confidence band
   if ConfChk.Checked then
   begin
-    ser := FChartFrame.PlotXY(ptLines, XPoints, LowConf, nil, nil, 'Lower confidence band', clRed);
+    SetLength(conf, ARegressionResults.Count);
+    for i := 0 to High(conf) do
+      conf[i] := ARegressionResults.ConfidenceLimits(XPoints[i], false);
+    ser := FChartFrame.PlotXY(ptLines, XPoints, conf, nil, nil, 'Lower confidence band', clRed);
     rightLabels.Add(ser.yValue[ser.Count-1], ser.YValue[ser.Count-1], 'LCL');
   end;
 
@@ -268,19 +211,26 @@ begin
 
   // Draw means
   if MeansChk.Checked then
-  begin
+    with ARegressionResults do
-    FChartFrame.VertLine(XMean, clGreen, psDashDot, 'Mean ' + XEdit.Text);
+    begin
-    topLabels.Add(XMean, XMean, 'Mean ' + XEdit.Text);
+      FChartFrame.VertLine(XMean, clGreen, psDashDot, 'Mean ' + XEdit.Text);
-    FChartFrame.HorLine(YMean, clGreen, psDash, 'Mean ' + YEdit.Text);
+      topLabels.Add(XMean, XMean, 'Mean ' + XEdit.Text);
-    rightLabels.Add(YMean, YMean, 'Mean ' + YEdit.Text);
+      FChartFrame.HorLine(YMean, clGreen, psDash, 'Mean ' + YEdit.Text);
-  end;
+      rightLabels.Add(YMean, YMean, 'Mean ' + YEdit.Text);
+    end;
 
   // Draw regression line
   if LineChk.Checked then
   begin
-    SetLength(tmpX, 2);  SetLength(tmpY, 2);
+    SetLength(tmpX, 2);
-    tmpX[0] := xmin;     tmpY[0] := tmpX[0] * slope + intercept;
+    SetLength(tmpY, 2);
-    tmpX[1] := xmax;     tmpY[1] := tmpX[1] * slope + intercept;
+    tmpX[0] := xmin;
+    tmpX[1] := xmax;
+    with ARegressionResults do
+    begin
+      tmpY[1] := tmpX[1] * Slope + Intercept;
+      tmpY[0] := tmpX[0] * Slope + Intercept;
+    end;
     ser := FChartFrame.PlotXY(ptLines, tmpX, tmpY, nil, nil, 'Predicted', clBlack);
     rightLabels.Add(tmpY[1], tmpY[1], 'Predicted');
   end;
@@ -289,67 +239,110 @@ begin
 end;
 
 
-procedure TPlotXYfrm.Reset;
+function TPlotXYForm.PrepareData(ADataGrid: TStringGrid;
+  out xCol, ycol: Integer; out XData, YData: DblDyneVec;
+  out ColNoSelected: IntDyneVec): Boolean;
+var
+  N: Integer;
+begin
+  Result := false;
+
+  ColNoSelected := nil;
+  XData := nil;
+  YData := nil;
+
+  xCol := GetVariableIndex(ADataGrid, XEdit.Text);
+  yCol := GetVariableIndex(ADataGrid, YEdit.Text);
+  if xCol = -1 then
+  begin
+    ErrorMsg('X variable not found.');
+    exit;
+  end;
+
+  if yCol = -1 then
+  begin
+    ErrorMsg('Y variable not found.');
+    exit;
+  end;
+
+  SetLength(ColNoSelected, 2);
+  ColNoSelected[0] := xCol;
+  ColNoSelected[1] := yCol;
+
+  XData := CollectValues(ADataGrid, xCol, colNoSelected);
+  YData := CollectValues(ADataGrid, ycol, colNoSelected);
+  N := Length(XData);
+  if N < 3 then
+  begin
+    ErrorMsg('At least three data points required.');
+    exit;
+  end;
+  if N <> Length(YData) then
+  begin
+    ErrorMsg('Equal count of cases required for x and y variables.');
+    exit;
+  end;
+
+  Result := true;
+end;
+
+
+procedure TPlotXYForm.Reset;
 var
   i: integer;
 begin
+  inherited;
+
   XEdit.Text := '';
   YEdit.Text := '';
   ConfEdit.Text := FormatFloat('0.0', DEFAULT_CONFIDENCE_LEVEL_PERCENT);
+
   LineChk.Checked := false;
   MeansChk.Checked := false;
   ConfChk.Checked := false;
+
   VarList.Items.Clear;
   for i := 1 to NoVariables do
     VarList.Items.Add(OS3MainFrm.DataGrid.Cells[i,0]);
 
-  if Assigned(FChartFrame) then
-    FChartFrame.Clear;
-  if Assigned(FReportFrame) then
-    FReportFrame.Clear;
-
   UpdateBtnStates;
 end;
 
 
-procedure TPlotXYFrm.UpdateBtnStates;
+procedure TPlotXYForm.UpdateBtnStates;
 begin
+  inherited;
+
   XInBtn.Enabled := (VarList.ItemIndex > -1) and (XEdit.Text = '');
   XoutBtn.Enabled := (XEdit.Text <> '');
 
   YinBtn.Enabled := (VarList.ItemIndex > -1) and (YEdit.Text = '');
   YoutBtn.Enabled := (YEdit.Text <> '');
-
-  if Assigned(FReportFrame) then
-    FReportFrame.UpdateBtnStates;
-  if Assigned(FChartFrame) then
-    FChartFrame.UpdateBtnStates;
 end;
 
 
-function TPlotXYFrm.Validate(out AMsg: String; out AControl: TWinControl;
+function TPlotXYForm.Validate(out AMsg: String; out AControl: TWinControl): Boolean;
-  Xcol, Ycol: Integer): Boolean;
 begin
   Result := false;
 
-  if (Xcol < 0) then
+  if XEdit.Text = '' then
   begin
-    AControl := XEdit;
+    AMsg := 'No variable selected for X.';
-    AMsg := 'No case selected for X.';
     exit;
   end;
 
-  if (Ycol < 0) then
+  if YEdit.Text = '' then
   begin
     AControl := YEdit;
-    AMsg := 'No case selected for Y.';
+    AMsg := 'No variable selected for Y.';
     exit;
   end;
+
   Result := true;
 end;
 
 
-procedure TPlotXYFrm.VarListDblClick(Sender: TObject);
+procedure TPlotXYForm.VarListDblClick(Sender: TObject);
 var
   index: integer;
 begin
@@ -366,13 +359,49 @@ begin
 end;
 
 
-procedure TPlotXYFrm.VarListSelectionChange(Sender: TObject; User: boolean);
+procedure TPlotXYForm.VarListSelectionChange(Sender: TObject; User: boolean);
 begin
   UpdateBtnStates;
 end;
 
 
-procedure TPlotXYFrm.XInBtnClick(Sender: TObject);
+procedure TPlotXYForm.WriteToReport(ARegressionResults: TBivariateRegressionResults);
+var
+  lReport: TStrings;
+begin
+  lReport := TStringList.Create;
+  try
+    lReport.Add('X vs. Y PLOT');
+    lReport.Add('');
+    lReport.Add('Data file: %s', [OS3MainFrm.FileNameEdit.Text]);
+    lReport.Add('');
+    lReport.Add('Variables:');
+    lReport.Add('  X: %s', [XEdit.Text]);
+    lReport.Add('  Y: %s', [YEdit.Text]);
+    lReport.Add('');
+    lReport.Add('Variable      Mean    Variance  Std.Dev.');
+    lReport.Add('----------  --------  --------  --------');
+    with ARegressionResults do
+    begin
+      lReport.Add('%-10s  %8.2f  %8.2f  %8.2f', [XEdit.Text, XMean, XVariance, XStdDev]);
+      lReport.Add('%-10s  %8.2f  %8.2f  %8.2f', [YEdit.Text, YMean, YVariance, YStdDev]);
+      lReport.Add('');
+      lReport.Add('Regression:');
+      lReport.Add('  Correlation:                %8.3f', [R]);
+      lReport.Add('  Slope:                      %8.3f', [Slope]);
+      lReport.Add('  Intercept:                  %8.3f', [Intercept]);
+      lReport.Add('  Standard Error of Estimate: %8.3f', [StdErrorPredicted]);
+      lReport.Add('  Number of good cases:       %8d',   [Count]);
+    end;
+
+    FReportFrame.DisplayReport(lReport);
+  finally
+    lReport.Free;
+  end;
+end;
+
+
+procedure TPlotXYForm.XInBtnClick(Sender: TObject);
 var
   index: integer;
 begin
@@ -386,7 +415,7 @@ begin
 end;
 
 
-procedure TPlotXYFrm.XOutBtnClick(Sender: TObject);
+procedure TPlotXYForm.XOutBtnClick(Sender: TObject);
 begin
   if XEdit.Text <> '' then
   begin
@@ -397,7 +426,7 @@ begin
 end;
 
 
-procedure TPlotXYFrm.YInBtnClick(Sender: TObject);
+procedure TPlotXYForm.YInBtnClick(Sender: TObject);
 var
   index: integer;
 begin
@@ -411,7 +440,7 @@ begin
 end;
 
 
-procedure TPlotXYFrm.YOutBtnClick(Sender: TObject);
+procedure TPlotXYForm.YOutBtnClick(Sender: TObject);
 begin
   if YEdit.Text <> '' then
   begin

applications/lazstats/source/forms/mainunit.lfm

@@ -19,7 +19,7 @@ object OS3MainFrm: TOS3MainFrm
   OnCreate = FormCreate
   OnDestroy = FormDestroy
   OnShow = FormShow
-  LCLVersion = '2.0.10.0'
+  LCLVersion = '2.1.0.0'
   object Panel1: TPanel
     Left = 0
     Height = 35

applications/lazstats/source/forms/mainunit.pas

@@ -2071,9 +2071,9 @@ end;
 // Menu "Analysis" > "Descriptive" > "Plot X vs Y"
 procedure TOS3MainFrm.mnuAnalysisDescr_PlotXvsYClick(Sender: TObject);
 begin
-  if PlotXYFrm = nil then
+  if PlotXYForm = nil then
-    Application.CreateForm(TPlotXYFrm, PlotXYFrm);
+    Application.CreateForm(TPlotXYForm, PlotXYForm);
-  PlotXYFrm.Show;
+  PlotXYForm.Show;
 end;
 
 procedure TOS3MainFrm.mnuAnalysisDescr_ResistanceLineClick(Sender: TObject);