facebookarchive / ml_sampler

Model assisted random sampling.
Other
121 stars 26 forks source link

Example: Best practices for repeated measurements #1

Open spencebeecher opened 7 years ago

spencebeecher commented 7 years ago

Outline best practices for how to understand if your ml sample is unstable (Score Weighted sampling + a model that overfits). Create a framework to illustrate 'expected' performance and outline how people can detect when their model (and sampling scheme) deviates from this.