I’m curious about the design choices behind UQLab. Specifically, I noticed that LARS (Least Angle Regression) is the default method for the coefficient calculation in PCE. Why was this method selected?
That’s an interesting question that @nluethen might be able to answer. In any case, you might be interested in her work titled “Sparse Polynomial Chaos Expansions: Literature Survey and Benchmark”: https://epubs.siam.org/doi/10.1137/20M1315774
Best regards
Styfen