Researchers of DiverSE have trained machine learning models to predict size or build status.
We should integrate them as part of the website.
There are several things to develop:
a predictor of size in which the user simply has to drop a .config
for each configuration of the database, add a new "size predictor" information that would call the predictor (we can also report the prediction errors by comparing the prediction with the actual size)
for each kernel version, report so-called "feature ranking list" using the machine learning models
for each kernel version, report so-called decision trees and feature importances about build status
Researchers of DiverSE have trained machine learning models to predict size or build status. We should integrate them as part of the website.
There are several things to develop: