Excuse my simple question but I couldn't locate how to test.
I've followed lastfm.py but couldn't find out at what point it uses existing training result. I am used to classical machine learning methods where you train a model then use this model when testing or evaluating some time later.
In tutorial_lastfm.ipynb, during recommendation phase, freshly trained model is being used. In examples/lastfm.py, calculate_similar_artists finds similarities and writes this to a file. I expected this file to be used in calculate_recommendations so the recommendations will be made based on the knowledge. Since I couldn't detect this part in the code I wanted to ask to find out what I am missing. For me it looks like calculate_recommendations just takes an algorithm name, not the existing file or model.
Excuse my simple question but I couldn't locate how to test.
I've followed lastfm.py but couldn't find out at what point it uses existing training result. I am used to classical machine learning methods where you train a model then use this model when testing or evaluating some time later. In tutorial_lastfm.ipynb, during recommendation phase, freshly trained model is being used. In examples/lastfm.py, calculate_similar_artists finds similarities and writes this to a file. I expected this file to be used in calculate_recommendations so the recommendations will be made based on the knowledge. Since I couldn't detect this part in the code I wanted to ask to find out what I am missing. For me it looks like calculate_recommendations just takes an algorithm name, not the existing file or model.