Preference Learning Toolbox
Preprocessing data
Before training the models, this screen allows the user to preprocess every and each of the features in the loaded dataset.
- Check-boxes: features can be removed from the dataset by unchecking the tick boxes in the left-most part of each line.
- Preprocessing buttons: this functionality includes a set of transformations that can be applied to a single feature (using the button in each row) or to all of them (using the button in the header). The following transformations are available:
- Binary transformation: given a nominal or ordinal feature, this transformation creates a binary encoding for each of the values of that feature. This encoding consists of a boolean feature for each distinct value; for every value, only the corresponding feature is true (1) while the others are false (0).
- Min-max transformation: the values for the selected features are transposed to fit a given value range (by default 0 and 1).
- Z-score transformation: the values of the selected feature are transformed so that the average value of the feature is zero and the standard deviation one.
- Data visualization: the right-most panel shows the values of the currently selected feature after the selected preprocessing method is applied.