kaggle-burn-cpu
Code for the Burn CPU, burn competition at Kaggle. Shows how to tune Extreme Learning Machines with hyperopt. Uses Python-ELM.
`elm.py, random_layer.py` - Python-ELM files - see `Python-ELM-LICENSE`
`data2num.py` - convert data to numbers only
`..._driver.py` - different flavours of optimization scripts to run
`hyperopt_logs` - results from each driver
`predict.py` - select best params from a log, train and save predictions
auxillary files for validation:
`bag_best.py` - select a few best models from a log, train and average their predictions
`rerun_best.py` - select a few best models from a log, re-run them to check the scores
`load_data.py` - a module for loading data used by `bag_best.py` and `rerun_best.py`
See http://fastml.com/extreme-learning-machines-and-optimizing-hyperparams-with-hyperopt/ for description.
License: BSD