Mateusz Wiza
- baseline.py - Script for applying single-target conformal regression with a single-targetregression model; All dependencies are given in requirements.txt; Original data set not provided due to confidentiality.
Contents:
- Data Preparation
- Experiment to determine the best performing underlying model
- Experiment to determine the best performing normalizing model
- Target-wise prediction interval width and validity visualization for optimal solution (can be reproduced using CSV files in 'graph-data' folder)
- multi-target-conformal-regression.py - Script for applying single-target conformal regression with a multi-target regression model; Additional dependency: nonconformist package modified to support multi-target underlying models (not included); All other dependencies are given in requirements.txt; Original data set not provided due to confidentiality.
Contents:
- Data Preparation
- Experiment to determine the best performing underlying multi-target model
- Target-wise prediction interval width and validity visualization for optimal solution (can be reproduced using CSV files in 'graph-data' folder)
- copula-based-conformal-regression.py - Script for plotting the results of copula-based conformal prediction. Results were generated by running a pre-existing algorithm (not included): https://github.com/M-Soundouss/CopulaConformalMTR/blob/master/code/conformal_multi_target_regression.py modified by replacing the underlying and normalizing models; Results are stored in JSON files (in 'graph-data/copulas' folder); All dependencies are given in requirements.txt;
Contents:
- Hyper-rectangle volume and validity visualization for optimal solution
- Target-wise prediction interval width visualization for optimal solution
This code was prepared in partial fulfilment of the requirements for the Degree of Bachelor of Science in Data Science and Artificial Intelligence, Maastricht University. Supervisors: Evgueni Smirnov, Ronald Frijns (Q-Park), Peter Steijns (Q-Park).