Get the latest tech news

The Effective Sample Size


Years ago I wrote about correcting covariate shift by reweighting your data. Your features come from the wrong distribution \(q\), you care about a target \(p\), so you weight every observation by \(\beta_i = p(x_i)/q(x_i)\) and your estimates are unbiased again.

None

Get the Android app

Or read this on Hacker News