acm_imbalanced_learning

This repo contains slides and code for the ACM Imbalanced Learning talk on 27th April 2016 in Austin, TX.

File listing

The files in the repo are listed below, with an explanation of what they're used for.

acm_imbalance_algorithms.ipynb - Jupyter notebook with scikit-learn classifiers training on the Kaggle dataset.
acm_imbalance_datasets.{pdf, pptx} - Powerpoint presentation with explanation of the dataset processing and algorithms.
acm_imbalance_sampling.ipynb - Jupyter notebook with a set of routines to pre-process imbalanced data.
acm_imbalanced_dataset.R - R script to use the 'unbalanced' package to pre-process data to remove imbalance.
datasets.zip - A zip file containing datasets for use in the talk. These are listed below
- cs-training.csv - Training data from the Kaggle 'Can I get some credit' competition
- cs-test.csv - Test data from the Kaggle competition.
- sampleEntry.csv - Sample entry format for the Kagle competition.
- cs-training-{CNN, OSS, smote, tomek}.csv - Processed training data (generated from acm_imbalanced_dataset.R) using the algorithms in the filename.

Any comments, questions, or feedback please submit a pull request !