A Hybrid Approach for Real Time Domains

Classification algorithms play a significant role in predicting the behavior of new data, based on the rules, which are extracted from the behavior of existing data in the database. This paper proposes optimal predictive approach with maximum accuracy and minimum risk factor involved. The main idea is to find best classification model for different real time domains by using a hybrid approach that is different from classical classification methodologies. Every classification data model has its accuracy measurement and error percentage or risk factor. We have focused on objective analysis of wrong prediction of these algorithms with some extended vision of including all possible groups of features. In other words our proposed approach facilitate the selection of most apt classification algorithm by adding an additional layer on classification model building process, in addition to data preprocessing step. The suitability of each classification algorithm is determined by optimal value analysis of algorithm accuracy and risk factor of accepting the wrong predictions as right ones.