Finite-State Extreme Effect Variable
Machine Learning
2020-01-01 v1 Information Retrieval
Machine Learning
Probability
Abstract
We generalize to the finite-state case the notion of the extreme effect variable that accumulates all the effect of a variant variable observed in changes of another variable . We conduct theoretical analysis and turn the problem of finding of an effect variable into a problem of a simultaneous decomposition of a set of distributions. The states of the extreme effect variable, on the one hand, are minimally affected by the variant variable and, on the other hand, are extremely different with respect to the observable variable . We apply our technique to online evaluation of a web search engine through A/B testing and show its utility.
Cite
@article{arxiv.1912.13377,
title = {Finite-State Extreme Effect Variable},
author = {Alexey Drutsa},
journal= {arXiv preprint arXiv:1912.13377},
year = {2020}
}
Comments
15 pages; 2 figures; 1 table