C# code of ECSM (AI 2016 conference)
- Download and install .NET Framework 4.0 (http://www.microsoft.com/en-us/download/details.aspx?id=17718).
- Run the program.
- Click Browse to select a dataset and a dictionary (optional).
- Specify the minimum difference support threshold (minDev).
- Click Mining to generate exceptional contrast sets. Select Save results if you want to save found exceptional contrast sets into csv file.
- Select Classification if you want to use exceptional contrast sets to classify test examples (we use 10-fold cross validation).
- The result includes mining time (in seconds), # of exceptional contrast sets, and accuracy, F1-score, Kappa statistic for classification.
In the source code, we use ALGLIB (http://www.alglib.net/) to compute Chi-square distribution. We also use Accord.NET Framework (http://accord-framework.net/) to compute F1-score and Kappa statistic.
If you use our source code, please cite our paper as follows:
Dang Nguyen, Wei Luo, Dinh Phung, Svetha Venkatesh (2016). Exceptional Contrast Set Mining: Moving Beyond the Deluge of the Obvious. AI 2016, Tasmania, Australia. Springer LNAI, 9992, 455-468