Conformal Predictor

Calculates the table with p-values for test data based on the provided calibration table. The p-value for a class is the fraction of entries from that class in the calibration set that have a model probability less than or equal to the model probability for the record under consideration. Small p-values indicate records that are nonconforming, larger p-values indicate records that are conforming. (Reference: Vovk, V., Gammerman, A. and Shafer, G., 2005. Algorithmic learning in a random world. Springer Science & Business Media.)

Input ports

1. Type: Table
   Prediction Table
   Table with predictions and class probabilities. Target variable must have the same domain as the calibration table.
2. Type: Table
   Calibration Table
   Calibration Table

Output ports

1. Type: Table
   Output
   Table with calibrated predictions and p-values.