IDENTIFYING THE REASON FOR MOBILE CALL DROPS USING DATA MINING TECHNOLOGY
ABSTRACT
Access to telecommunication services is indispensable for the comprehensive development of a nation’s economy, spanning sectors such as manufacturing, banking, education, agriculture, and government. However, telecommunication services are not immune to challenges, and one of the foremost issues faced by telecom operators is mobile call drops. In the realm of telecommunications, mobile call drop occurs when calls are prematurely terminated before both parties have concluded their conversation or before either party has intentionally ended the call. This research endeavors to create a predictive model capable of identifying instances of mobile calls. To surmount the limitations of simplistic statistical approaches, data mining techniques, methodologies, and methods are proposed and employed in this study. Approximately 20,000 records spanning one and a half years of Fault Management data were selected for analysis. Irrelevant and extraneous data were meticulously filtered out, resulting in a dataset comprising 16,996 records with eight attributes, forming the basis of this study. Data preprocessing was conducted to cleanse and refine the dataset, making it suitable for data mining tasks. The study was executed using the WEKA software version 3.8.1, employing four classification techniques: J48 and Random Forest algorithms from the decision tree family, as well as PART and JRIP algorithms from rule induction. The results reveal that the J48 decision tree algorithm, validated through 10-fold cross-validation, achieved superior performance with an accuracy rate of 95.43% and rapid processing speed of 0.06 seconds. This algorithm effectively employed eight attributes for the research’s experimentation. Challenges encountered during the study included the absence of relevant prior work in the field of telecommunication mobile call drop analysis and the limitation that the FMS mobile network server could only retain two years’ worth of data due to its substantial volume. Consequently, data preceding the two-year limit could not be obtained
IDENTIFYING THE REASON FOR MOBILE CALL DROPS USING DATA MINING TECHNOLOGY, GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS