ML models often mispredict, and it is hard to tell when and why. We present a data mining based approach to discover whether there is a certain form of data that particular causes the model to mispredict. - View it on GitHub