Benchmark for Generalized Linear Models (GLM)

Benchopt is a package to simplify and make more transparent and reproducible the comparisons of optimization algorithms. The GLM consists in solving the following program:

$\min_{w} f(W^\top X, y) + \lambda R(w)$

where $f$ is a link function, defined by the distribution of the target variables, and $R$ is a regularisation function -- typically the $\ell_1$ or $\ell_2$ norm.

Install

This benchmark can be run using the following commands:

$ pip install -U benchopt
$ git clone https://github.com/benchopt/benchmark_glm
$ benchopt run benchmark_glm

Apart from the problem, options can be passed to benchopt run, to restrict the benchmarks to some solvers or datasets, e.g.:

$ benchopt run benchmark_glm -s solver1 -d dataset2 --max-runs 10 --n-repetitions 10

Use benchopt run -h for more details about these options, or visit https://benchopt.github.io/api.html.