Predicting the genre of songs for given song ids. User song listening activity history and song duration related data is provided for learning.
Provided Problem Statement:
Predict the genre of some tracks with following available data: session.csv: This file contains records of various user session listening specific tracks of their choice at different point of time. More specifically each record has set of three field user_id, song_id and timestamp.
tracks.csv: It has records of different tracks with their time duration and genre. Each record consists of three field song_id, duration and genre respectively.
tracks_to_complete.csv: This file contains test data having song_id for which genre needs to be predicted.
Execution Environment & Required Library:
- Python 2.x
- Libraries
- Scikit Learn
- Numpy
- Pandas
- Matplotlib
Solution:
- solution.py - Development code for data preprocesing, feature engineering to create worthful fetures, ML modeling and finally validation and results.
- solution.pdf - Transforms problem statements to approach it effectively, then explains all mandatory steps and possible reasoning behind feature engineerings & model selection. explains further results and justifies the model performance.
Note for other Enthusiastic Contributor:
Most welcome to extend & contribute this experiment with better feature engineering & ML modeling.
Queries?...Connect with me at: