dask-glm

dask-glm copied to clipboard

SVD solver

1

Hi, In my current project, my team needs to use RidgeRegression with an SVD solver. I implemented an initial version [here](https://github.com/mathdugre/dask-glm/commit/bebf23b5040426da3755bf2c3032a871a165fec5?diff=split), based on the [sklearn implementation](https://github.com/scikit-learn/scikit-learn/commit/814ad9ba14ea1f53da353368d34daf89061cf92e#diff-e7a1a6bac747c5273cc1e858fb418e7a86ceb453dc8a9ba92839da3d69932ba9). I also made an...

mathdugre

Remove estimators and update CI build/release

1

milesgranger

Verify that Read the Docs is building after master -> main

4

Spotted on https://github.com/dask/dask-cloudprovider/issues/244#issuecomment-795310925. Read The Docs was still building `latest` from the `master` branch and builds were failing. It was configured to build from the default branch, but had cached...

jsignell

Warning on config "optimization.fuse.ave-width" w/ solver="lbfgs"

1

**What happened**: When I set `solver=lbfgs`, I get a warning: > /Users/scott/anaconda3/lib/python3.8/site-packages/dask/config.py:588: UserWarning: Configuration key "fuse_ave_width" has been deprecated. Please use "optimization.fuse.ave-width" instead **What you expected to happen**: I did...

stsievert

setup-miniconda v2 fixes some deprecations in GitHub Actions

jacobtomlinson

[WIP] Allow pure numpy array (not dask array) as inputs

2

Currently `dask_glm.estimators` only accepts `dask.array` as inputs due to the line below and other places where `._meta` is accessed without checking the data type. https://github.com/dask/dask-glm/blob/7b2f85fe043eb29212755e67e33e3df553ed0e58/dask_glm/estimators.py#L67 https://github.com/dask/dask-glm/blob/7b2f85fe043eb29212755e67e33e3df553ed0e58/dask_glm/utils.py#L120-L124 Click to see the...

daxiongshu

Add n_iter_ attribute to estimators

5

After this is merged, Dask-ML must be adapted to this change as well. https://github.com/dask/dask-ml/blob/master/dask_ml/linear_model/glm.py#L187 Solves #77.

pentschev

[WIP] Adjust all algorithms to work with CuPy

8

Requirements: * https://github.com/numpy/numpy/pull/13046 * https://github.com/cupy/cupy/pull/2079 Resolves #73.

pentschev

change to Numpy power ufunc

3

Change builtin python power function to Numpy power ufunc

edenbaus

WIP: ENH: Basic implementation of SGD

14

This implements the popular mini-batch stochastic gradient descent in dask-glm. At each iteration, it grabs `batch_size` examples, computes the gradients for these examples, and then updates the parameter accordingly. The...

stsievert