kjahan / evaluation

Recommendations, ranking, and search metrics

evaluation

Recommendations, ranking, and search metrics

Setup environment

$ conda create -n recom python=3.6

$ conda install -n recom pandas

$ conda install -n recom numpy

$ conda install -n recom pytest

pip install implicit==0.4.0

conda install -c conda-forge notebook -n recom

conda install nb_conda -n recom

conda install h5py -n recom

conda install -c conda-forge ipywidgets -n recom

jupyter nbextension enable --py widgetsnbextension

Activate conda env:

$ conda activate recom

(On MacOS: unset PYTHONPATH)

Deactivae conda env:

$ conda deactivate

Run tests:

pytest

ALS with MovieLens Demo:

To run ALS with Movie Lense data and get offline metrics (Precison@k & Recall@k), run:

python src/implicit_demo.py