APES

Exhaustive variable selection is known to be time consuming, especially for Generalised Linear Models (GLMs). APES is a variable selection method that first converts a given GLM into a linear model first and then uses a best-subset algorithm to find the best linear model. This linear model is then converted back to a GLM to approximate the original exhaustive search problem. APES can be orders of magnitudes faster than the true exhaustive search while retaining a reasonable accuracy.

Kevin Y.X. Wang
Data Scientist

Data scientist at Illumina. PhD in Statistics.