Measure Effectiveness of SMS Spam Detection Model Based on Machine Learning Techniques

Abstract

With the increase in the use of mobile phones, the use of Short Message Service has increased exponentially. With the cost of text messages dropping, people started using them for promotional purposes and unethical activities. This led to a massive increase in spam and consequently the loss of personal and financial data. To prevent data loss, it is essential that spam is detected as quickly as possible. Thus, this paper aims to classify spam not only effectively but also in a short time using python. A dataset containing thousands of text messages containing natural messages (ham) and spam messages was used. Natural language processing techniques were used Multiomail Naive Bayes, Decision Tree and Random Forest are used through which we can classify the message type. After applying these algorithms, Random Forest algorithm got the best accuracy 0.99% in 0.15 second.