- Improve weights to combine Linear+SVM+Choudhary
- Implement k-fold cross-validation
- Improve the SVM (feature definition, tuning, &c)
- Look more closely at raw data
- Look more closely at which outliers should be dropped
- Transform skewed variables
- Look at interaction terms
- Tune parameters
- Find a way to fold in results from non-linear fits
- Start over in Python
- Reorganize this repository so top level not so crowded
- Revise text of Kaggle notebook kernel
- Dowload and push latest Kaggle notebook kernel to here
Notebook-ify Amit Choudhary's Python script