A Prediction System for Bike Sharing Using Artificial Immune System with Regression Trees

In past years, AIS are powerful and useful algorithms to solve classification and optimal problems such as intrusion detection, scheduling and parameters optimization. However, AIS has rarely been applied in solving the prediction problem. In this paper, we propose a novel model by combining AIS with regression trees (RT) prediction system for a real world application, i.e., A bike sharing system (BSS). The cells in AIS are the basic constituent elements and we embed RT forecasting sub-models in the AIS to form cells pool and use clone selection mechanism to generate cloned antibody. Therefore, AIS-RT prediction system can be applied to solve the prediction problem. Experiments have been conducted for AIS-RT on bike sharing system. Experimental results show that the AIS prediction system can further improve the performance of an adopted forecasting model, and furthermore outperform the performances of other ensemble approaches.