PROJECT TITLE :

DPPred: An Effective Prediction Framework with Concise Discriminative Patterns Sign In or Purchase - 2018

ABSTRACT:

In the literature, two series of models are proposed to deal with prediction issues together with classification and regression. Simple models, such as generalized linear models, have standard performance but robust interpretability on a group of easy options. The other series, together with tree-based models, organize numerical, categorical, and high dimensional options into a comprehensive structure with wealthy interpretable information in the info. During this Project, we have a tendency to propose a novel Discriminative Pattern-based Prediction framework (DPPred) to accomplish the prediction tasks by taking their advantages of both effectiveness and interpretability. Specifically, DPPred adopts the concise discriminative patterns that are on the prefix paths from the foundation to leaf nodes within the tree-based models. DPPred selects a restricted variety of the useful discriminative patterns by searching for the foremost effective pattern combination to suit generalized linear models. Intensive experiments show that in many scenarios, DPPred provides competitive accuracy with the state-of-the-art and the dear interpretability for developers and specialists. In particular, taking a clinical application dataset as a case study, our DPPred outperforms the baselines by using solely forty concise discriminative patterns out of a potentially exponentially massive set of patterns.