Imbalanced Learning with Parametric Linear Programming Support Vector Machine for Weather Data Application

Abstract

Imbalanced learning is a challenging task in predictive modeling and machine learning that has inspired many researchers to attempt to improve the existing algorithms for more accurate predictions of imbalanced data sets. Due to the nature of rare events, developing reliable and efficient classification models for imbalanced data has not been easily accomplished, and over the past two decades, various methods have been proposed. To this end, we propose a Linear Programming Support Vector Machine (LP-SVM) model to address the issue of imbalanced learning in weather applications. To further improve the model’s predictive accuracy, we have implemented a parameter selection method based on the multi-objective parametric simplex approach for parameter tuning of LP-SVM. For numerical tests, we have used a real data set consisting of weather observations made by the Bureau of Meteorology’s (BM) system in Australia. The results obtained from training and testing the model demonstrate the effectiveness of our proposed model on the tested examples.