
demo_waid.exe
The WAID Visual Basic demonstration
This demonstration provides an
interface to the Weighted Automatic Interaction Detection (WAID) software,
see EUREDIT D4.2.1 and D4.2.2 by Xinqiang and Chambers (2002).
The demonstration is used as follows:
- Select the XML data file by using the ListBoxes provided in the upper half of the form.
- Load the data by clicking the "Load Data" button.
- Click on the "Set Options" button, and a dialog with a range of options will appear.
The available options are as follows:
- IMPUTATION METHOD: The following options are available
- Mean: specifies that the mean of the data in a leaf node is to be used.
- Mode: specifies that the mode of the data in a leaf node is to be used.
- Random Selection: specifies that a random value from the data in a leaf node is to be used.
- WEIGHT SCHEME: The following options are available
- Least Squares: specifies ordinary least squares and each weight is set to 1
- Andrew Sine: specifies the Andrew Sine weight function
- Tukey: specifies the Tukey bi-weight function
- Huber: specifies the Huber weight function
- Tuning constant C: specifies the value of the constant to be used in the Andrew Sine, Tukey and Huber weight functions.
- Maximum number of iterations: specifies the maximum number of iterations that are to be used in the calculation of the weights.
- Variables and Selected variables: All the variables in the data file are shown in the left hand ListBox.
The selected variables are shown in the right hand ListBox. Variables can be added to the right hand ListBox by using the mouse
to select variables from the the left hand list and then clicking on the button labelled ">>".
Variables can be removed from the right hand list by selecting them with the mouse and using
the button labelled "<<".
- Variable to impute: this is set by doubling clicking on the variable in the left hand ListBox.
The demonstration currently only processes a single selected missing data pattern at a time, and the selected data pattern can only contain one missing (dependent) variable
per record. If the data (as is usual) contains more than one missing variable per record then it is necessary to process each missing variable
separately.
- Click on the button labelled "Load Options" to store the selected data processing options.
- To process the data click the button labelled "WAID"
- The processed data can be viewed by clicking on the "View Data" button.
- If more processing is required, then go to step 3.
Platform software homepage