Browsing by Author "Mahanama, Thilini"

Now showing 1 - 2 of 2

Defaulter Prediction in the Fixed-line Telecommunication Sector Using Machine Learning
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Ginige, Sachini; Rajapakse, Chathura; Asanka, Dinesh; Mahanama, Thilini
In the modern connected era, the telecommunications sector plays a critical role in enabling efficient business operations across all industries. However, defaulting customers who fail to pay their dues after consuming services remain a significant challenge in the industry. Defaulters pose a risk to service providers, calling for measures to lessen both the probability of occurrence as well as its impact. Early identification of defaulters through prediction is a possible solution that enables proactive measures to mitigate the risk. However, the nature of the fixed-line product segment poses additional constraints in identifying defaulters, highlighting an existing knowledge gap. The research aims to evaluate the effectiveness of machine learning as a technique for the prediction of defaulters in the fixed-line telecommunication sector, and to develop an effective predictive model for the purpose. The success of machine learning techniques in analysis and prediction over traditional methods prompted its use in this study. The study followed the design science research methodology. An analysis was conducted based on past transaction data. Special consideration was given to the scenario of customers with little to no transaction history. Based on the analysis, a feature list for identifying defaulters was compiled, and multiple predictive models were developed and evaluated in comparison. The resulting predictive model, which uses the Random Forest technique, shows high performance in all considered aspects. The findings of the study demonstrate that machine learning techniques can effectively predict defaulters in the fixed-line telecommunication sector, with significant implications for mitigating the risk associated.
Predicting the Severity of Tornado Events by Learning a Statistical Manifold for Tornado Property Losses
(Journal of Environmental Accounting and Management, 2023) Mahanama, Thilini; Paranamana, Pushpi; Volchenkov, Dimitri
We examine the relationship between property losses caused by tornadoes and their physical parameters, namely the tornado path length and width, using data reported by the National Oceanic and Atmospheric Administration in the United States. We observe that the statistics of property losses cannot be described by a single distribution but rather by a two-dimensional statistical manifold of distributions that may re ect two di erent mechanisms of property loss compensations. Assessing the di erence between distributions of losses caused by tornadoes using Kolmogorov-Smirnov's distance, we construct the 2-D manifold using the method of multi-dimensional scaling. Then we de ne a curvature coe cient that characterizes the contraction and expansion of the derived manifold to explain the complex dynamics of the probability distributions of losses. The regions with expansions identify the ranges of physical parameters for which the extreme tornado events may occur, which helps in assessing compensation strategies.