It may be possible to automate some datascience parts ?
- Technical feature engineering
- Grid search using hyperopt
- Basic stacking
This is a modest try to do so begining with hyperoptimisation of datascience models
You will need xgboost, hyperopt and the latest Anaconda distribution (for scikit-learn, pandas, numpy and scipy)