skglm
skglm copied to clipboard
scikit-learn-contrib
Fast and modular sklearn replacement for generalized linear models
## Description of the feature Handle $\log\det $-like datafit and allow variables with multiple roles. **Additional context** @bgirault-inria is currently solving (simpler) graphical-Lasso-like problems and could potentially benefit from our...
Reproduce with ```python import numpy as np from skglm.utils.jit_compilation import compiled_clone from skglm import datafits from skglm import penalties from skglm.solvers import ProxNewton from skglm.utils.data import make_correlated_data X, y, _...
Popped up in #251 and several other times. We have penalty.alpha_max for some penalties; having a function which does not require the gradient, but takes datafit, penalty, X and y...
As highlighted in #256, `boundscheck` are disabled in numba compiled functions for the sake of performance. However, this might hide tough bugs. It is possible to enable `boundscheck` globally using...
Because the parameters that we would like to cross validate are parameters of model.penalty, model.datafit or model.solver, we are not comaptible: ```python from skglm.utils.data import make_correlated_data from skglm.datafits import Quadratic...
A datafit can implement a score (based on its use for classif or regression) and this score can be accessed by GeneralizedLinearEstimator.score() This would be one more step towards sklearn-like...
## Description of the feature **Exact feature** Solve the following optimization problem $$\arg \min_{\beta} \frac{1}{2n} \|| X \beta \||^2 + \frac{1}{n} \beta^\top X^\top y + \text{penalty} \enspace,$$ with no access...
@Badr-MOUFAD there is this comment > # Despite violating the conditions mentioned in [1] # this choice of steps yield in practice a convergent algorithm # with better speed of...
Add a non convex graphical lasso solver with reweighting: - use sklearn code to implement a WeightedGraphicalLasso - add a wrapper to fit it ~5 times a la IterativeRewieghtedL1 with...
Metadata
Owner
Metadata
Fast and modular sklearn replacement for generalized linear models