* [train][datasets] add example for big data training * add title docstring * lint and dependencies * add dask_ml requirement