Skip to content
Playlist recommendation
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore
LICENSE.md
README.md
album.py
artist.py
data_frame_util.py
date_time_util.py
info.py
logger.py
model_util.py
playlist.py
playlist_parser.py
playlist_slice.py
playlist_slice_converter.py
playlist_util.py
ranging_matrix_factory.py
track.py
track_filter.py
uthoern.py
uthoern_predict.py

README.md

Uthoern - TeamLuebeck

Playlist continuation recommender system.

Installing

The only required installation step is to install Anaconda. Uthoern needs libaries like Pandas or SciKit Learn and there are included in the Anaconda package

Execution

  1. Go into the project
  2. Create there a folder named 'model_storage'
  3. Open the Anaconda Prompt
  4. Navigate into the project folder

Training

  1. Replace <Path_to_mpd_folder> with the path to the Million Playlist Dataset and execute the statement:
python3 uthoern.py <Path_to_mpd_folder> 1000
  1. The Training needs around 30 hours. If you want to see the current state, lock into the utheorn.log file inside the project.

Prediction

  1. Now enter the following statement. <Path_to_the_challenge_set_folder> is the path to the challenge set. The TrainingId looks like 20180627114113583448. You will find your id in the model_storage folder.
python3 uthoern_predict.py <Path_to_the_challenge_set_folder> <TrainingId>

This steps needs around 40h

Prediction

If the prediction process ist finished, you will find your playlist recommendation in the 'model_storage/TrainingId' folder named TrainingId_TeamLuebeck__mdp_submission.csv

You can’t perform that action at this time.