Application of Machine Learning in Predicting Call Quality of Telecom Service Providers

In today’s digital era mobile phones have become a necessity and important part of every one’s life. The call quality is a developing research area in the field of cellular networking domain. The quality of call for a set of calls is predicted based on the aggregation of successful parameters of a call. The telecom operators face a major issue in their service which is referred as call drop. Call drop is a situation in which the ongoing conversation of a subscriber is ended in between abruptly. The unexpected drop in a call is becoming one of the biggest problems in telecom sector globally. The mobile call drop affects the call quality of subscribers which leads to reduction in revenue among telecom service. This research aims to design a machine learning model in predicting Call Quality of Telecom service providers. To overcome the call drop issue, this research has used a model of neural network to improve their performance and effectiveness of call. This research was conducted using Python software and Random Forest Classifier (machine learning algorithm) is used. The call parameters used in this research are call data, call rating, number of subscribers and call drop category. The main idea of the research is to develop a machine learning model to estimate the parameters of call during a call and presents an average rank at the termination of a call. This research conducts the research for three months only namely September, October and November month. The performance is evaluated in this research using accuracy, precision, recall, and f-score measures. The results of the research render an accuracy of 90% for call quality. This research suggests telecom operators to use machine learning technique on cellular network data for reducing the call drops and resolve it to improve subscribers call quality for a non-stop conversation in future.