Create functions to compare various algorithms and parameterizations

rademacher-p / stats-learn

Python package for statistical learning with a Bayesian focus

https://rademacher-p.github.io/stats-learn/

MIT License

1 stars 0 forks source link

Create functions to compare various algorithms and parameterizations #12

Closed rademacher-p closed 3 years ago

rademacher-p commented 4 years ago

[x] For different N
[x] For different parameterizations of single learner
[x] For various arbitrary learners
[x] SE bias/variance visualization
[x] Use warm_start functionality for efficient computation

rademacher-p commented 3 years ago

For assessing risk for fixed N on a single learner for varying parameterization, it is desirable to use the same predictor object and simply set an attribute to a new value before calling predict. No new fit call should be needed and same data can be used.