DTREG COM Library

The DTREG COM (Component Object Model) library makes it easy for production applications to call DTREG as an “engine” to compute the predicted value for data records using a decision tree model. You must use the GUI version of DTREG to construct a model before you can use it with the DTREG COM library to predict values.

Any type of model (Single Tree, TreeBoost, Decision Tree Forest, Multilayer Perceptron Neural Networks Cascade Correlation Neural Networks Probabilistic and General Regression Neural Networks (PNN/GRNN) Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), or Logistic Regression) can be used with the DTREG COM library to generate predicted values. All of the advanced scoring features such as the use of surrogate splitters to handle missing predictor values are used in the DTREG COM library.

Because of the standardization of the COM interface, it is easy to call the DTREG COM library from programs written in Visual Basic, Visual C++, VBA, Excel, Access, ASP and other languages. The DTREG COM library is designed to run as an in-process DLL for speed of execution.

Click here to view the manual for the DTREG COM Library

Example Visual Basic Program That Calls The DTREG COM Library

Private Sub RunTest_Click()
'
'  Reference the DTREG COM library.
'
Dim dtreg As DTREGCOMLib.dtreg
Set dtreg = New DTREGCOMLib.dtreg
'
'  Miscellaneous variable declarations.
'
Dim ProjectFile As String
Dim ModelType, status, index As Long
Dim NumVar, NumCat As Long
Dim VarClass, VarType As Long
Dim VarName, CatLabel As String
Dim ixSepalLength, ixSepalWidth As Long
Dim ixPetalLength, ixPetalWidth As Long
Dim ixSpecies As Long
Dim PredictedClass As String
Dim CatProb As Double
'
'  Open the DTREG project file (TreeBoostIris.dtr).
'
ProjectFile = "c:\DTREG\Test\TreeBoostIris.dtr"
status = dtreg.OpenProjectFile(ProjectFile)
If (status <> 0) Then
    boxStatus = "Error opening project file: " + Format(status, "#")
    Stop
End If
'
'  Find out what type of model this is
'
ModelType = dtreg.ModelType
'
'  Find out how many variables are in the model.
'
NumVar = dtreg.NumberOfVariables
'
'  Check the name and properties of each variable.
'
For index = 0 To NumVar - 1
    VarName = dtreg.VariableName(index)
    VarClass = dtreg.VariableClass(index)
    VarType = dtreg.VariableType(index)
Next
'
'  Get the index numbers of the variables variables.
'
ixSpecies = dtreg.VariableIndex("Species")
ixSepalLength = dtreg.VariableIndex("Sepal length")
ixSepalWidth = dtreg.VariableIndex("Sepal width")
ixPetalLength = dtreg.VariableIndex("Petal length")
ixPetalWidth = dtreg.VariableIndex("Petal width")
'
'  Set the values of the predictors we want to score.
'
status = dtreg.SetVariableValue(ixSepalLength, 5.1)
status = dtreg.SetVariableValue(ixSepalWidth, 3.5)
status = dtreg.SetVariableValue(ixPetalLength, 1.4)
status = dtreg.SetVariableValue(ixPetalWidth, 0.2)
'
'  Compute the predicted target category.
'
PredictedClass = dtreg.PredictedTargetCategory
'
'  See if any error occurred during the computation.
'
status = dtreg.LastStatus
If status <> 0 Then
    boxStatus = "Error computing target: " + Format(status, "#")
    Stop
End If
'
'  If we are using a TreeBoost model, check the probabilities
'  for each of the target variable categories.
'
If ModelType = 2 Then
    NumCat = dtreg.NumberOfCategories(ixSpecies)
    For index = 0 To NumCat - 1
        CatLabel = dtreg.CategoryLabel(ixSpecies, index)
        CatProb = dtreg.TargetCategoryProbability(index)
    Next
End If

End Sub