Comparison of Feature Selection and Classification Algorithms for Restaurant Dataset Classification

Currently, the rapid growth of information on the Internet makes automatic text classification play an important role to help people discovering desired information on enormous resources. Text mining, feature selection and classification algorithm have effect on the classification performance directly. In this paper, the comparative study of the text classification performance is proposed. It compares between three feature selection approaches with four classification algorithms. The algorithms include C4.5, ID3, Bayes Net and Naïve Bayes. The objective is to find suitable model for restaurant dataset. The experimental results indicated that when a set of selected features was large, C4.5 and ID3 algorithms generated models with better accuracy than those induced by Bayes Net and Naïve Bayes algorithms. On the other hand, the small amount of selected features makes every algorithm yields models with almost the same accuracy. However, algorithms based on Bayes’ theorem could generally induce classification models with higher accuracy than the tree-based algorithms. Key-Words: Classification Algorithms, Feature Selection, Restaurant Classification, Text Classification