Using machine learning models to predict the consumer behavior in telecommunication

Abstract

Researching and applying machine learning models in real-life domains tend to attract communities such as scientists and businesses. This paper describes the results of our analysis of applying machine learning models to the field of telecommunications. Specifically, in our research, we used classification models to predict consumer behavior and performed evaluation on 10 different machine learning models based on F1-Score, Precision, Recall, and Accuracy. Experiments have been carried out on 6000 different samples collected from the Center of Information Technology, VNPT Hai Duong. We found that using machine learning models for the problem is a very promising approach to predict consumer behaviors in telecommunication. In particular, the Gradient Boosting model provides a very high recall – 0.986. This significant result is the premise to improve and develop new models with higher efficiency in the future for businesses to make more suitable business decisions for each type of customers.