Predictive Systems Calibrator (Regression)

Creates a calibration table with ranking for further calibration of the test data sets. Ranking is a sorting by absolute error (Alpha) provided by predictor nodes in descending order. In order to use normalization Prediction Variance or estimate of difficulty - Sigma (needs to be calculated separately) should be provided. Alpha values are used for further cumulative distribution function (CDF) calculation. This node is supposed to be used WITH Conformal Calibration and Conformal Prediction loops.

Input ports

1. Type: Table
   Input
   Table provided by predictor nodes fitted predictions.

Output ports

1. Type: Table
   Output
   Calibration table with ranks for each sample.