Participated in a DrivenData competition
At the time of writing this, the model is hodling place 42 on the leaderboard.
The main goal was to learn about Sklearn's machine learning/label prediction capabilities.
The process:
- data parsing & formatting & cleaning & transforming
- feature extraction
- trainingset splits
- model training
- implement CrossValidation
- implement HyperParameter Optimization
- model persistence (pickle)
- simplify multilabelling & text merging functions