Fuzzy case-based reasoning for weather prediction

Abstract

Prediction is the process of the estimation of unknown situation that refers to time-series, cross sectional or longitudinal data. Weather prediction is the process to project how to atmosphere will evolve. Weather is known as continuous, data-intensive, multidimensional, dynamic and chaotic. The chaotic nature of atmosphere required the massive computational power in order to solve the equations that describe the atmosphere, and the incomplete understanding of weather can make the prediction become less accurate. Based on this problem, Fuzzy Case-Based Reasoning (FCBR) is introduced in solving the prediction problem. Fuzzy can have the degree of truthfulness and falsehood that can handle uncertainty of the chaotic variables of the weather. Meanwhile Case Based Reasoning (CBR) has an ability to identify the similar cases from the past using the similarity measurement technique such as Euclidean distance. CBR can reduce the knowledge acquisition task and can reason with incomplete and imprecise data or knowledge. This study is conducted to investigate how fuzzy and CBR could solve the prediction problem and how it can improve its performance. From the experiment, it shows that the Fuzzy Case Based Reasoning has improved the accuracy of the weather prediction with achievement of 87%.