2019/02/11 by Iqbal H. Sarker, Sarker, Iqbal H. · 1 citation
Computer Science · Social Sciences · #Computers and Society (cs.CY) #Data Management and Algorithms #Data Stream Mining Techniques #FOS: Computer and information sciences #Human Mobility and Location-Based Analysis #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Recommender Systems and Techniques #Social and Information Networks (cs.SI) #Web Data Mining and Analysis
paper · pdf · doi:10.48550/arxiv.1902.07588
openalex publication_date 2019/02/11 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28
Real-life mobile phone data may contain noisy instances, which is a\nfundamental issue for building a prediction model with many potential negative\nconsequences. The complexity of the inferred model may increase, may arise\noverfitting problem, and thereby the overall prediction accuracy of the model\nmay decrease. In this paper, we address these issues and present a robust\nprediction model for real-life mobile phone data of individual users, in order\nto improve the prediction accuracy of the model. In our robust model, we first\neffectively identify and eliminate the noisy instances from the training\ndataset by determining a dynamic noise threshold using naive Bayes classifier\nand laplace estimator, which may differ from user-to-user according to their\nunique behavioral patterns. After that, we employ the most popular rule-based\nmachine learning classification technique, i.e., decision tree, on the\nnoise-free quality dataset to build the prediction model. Experimental results\non the real-life mobile phone datasets (e.g., phone call log) of individual\nmobile phone users, show the effectiveness of our robust model in terms of\nprecision, recall and f-measure.\n