All submissions made to Kaggle are located in the submission
folder. Results for model tuning are located the intermediates
folder and both raw and processed data are located in the data
folder.
Two scripts, clean.R
and clean_test.R
, handle data wrangling for the training and testing set, while explore_and_recipe.R
contains code for all recipes.
Training scripts are named like job_<modelType>.R
, while model selection, prediction and csv writing are handled in evaluate_<modelType>.R
To run the full pipeline, scripts should be run in the following order.
-
clean.R
andclean_test.R
-
explore_and_recipe.R
-
job_bt.R
,job_rf.R
-
evaluate_bt.R
andevaluate_rf.R
-
job_ensemble.R
andevaluate_ensemble.R