Advanced analytics, big data, predictive modelling, deep learning

The purpose of testing analysis in predictive analytics is to compare the response of a predictive model against the actual target values in an independent testing set. There are several techniques that can be used for testing the performance of a predictive model. Receiver operating characteristic (ROC) curve is one of the most useful testing methods for binary classification problems, since it provides a comprehensive and visually attractive way to summarize the accuracy of predictions. By moving the decision threshold, we change the number of instances classified as positives and negatives. If the score of an instance is greater or equal to the threshold, then it will be classified as positive.