README.md

Mastering Uncertainty in Performance Estimations of Configurable Software Systems

Supplementary Material

With this web site, we provide the following material:

• descriptions of the subject systems we used
• additional results of our accuracy evaluation in RQ1 & RQ2 including the absolute model
• more perspective on the uncertainty calibration of our models

We also provide data containing the accuracies of our models as well as predicted confidence interals as well as an implementation of our probabilistic programming approach with an updated Dockerscript for an easier set-up.

Extended Supplementary Material

As a result of a more detailed analysis uncertainty within the influences of options and interactions (terms) on execution time and energy consumption, we additionally provide

• term influence confidence interval widths for models trained on T₂ vs models trained on T₃
• prediction confidence interval widths for models trained on T₂ vs models trained on T₃

as well as

• a detailed analysis of the time needed to fit models with our approach