This project attempts to predict whether a user is interested in listening to music at current time t, given their past history. The project makes use of Scikit-Learn and Tensorflow.
The model makes usse of the LastFM 100k user dataset which was transformed into time-series data and stored in SQLite.
Methods applied were:
- Logistic regression
- Linear SVM Model
- RBF SVM Model
- RNN-LSTM
Shortcut to the final report is here
- 1_codemodule folder - Some common code
- 2_Settings - Stores database and file paths
- 3_Data - Not in git as too large
- 4_DataProcessing - Transformation of raw data into time-series data
- 5_PreliminaryAnalysis - Quick look at the raw and transformed data
- 6_MainModel - Main code