A hybrid approach for feature selection based on global and local optimization for email spam detection

These days, Due to the universal availability of internet connections, email correspondence has become one of the most cost-effective and efficient approaches for official and business users. Millions of spam emails are being shared every day. To protect individuals or organizational data, Spam detection is needed. In Spam detection, handling high dimensional datasets in machine learning is high in time complexity and space complexity. To decrease time complexity and space complexity, feature selection is required to remove the irrelevant features. In this paper, the proposed idea is to optimize the feature selection method to decrease time complexity, space complexity and increase accuracy as well. The proposed feature selection method is a hybrid method based upon two optimization techniques; the first one is global optimization using the chi2 selectkbest method, The second one is local optimization using the Tree-based feature selection method. Experiments are carried out to compare four different classifiers. The proposed idea performs well on precision, memory, and accuracy efficiency tests, according to the findings.