Association rule mining using time series data for Malaysia climate variability prediction

Abstract

Many studies have been conducted to determine how data mining can be used in predicting climate change. Previous studies showed many data mining methods have been used in related to climate prediction, however classification and clustering methods are widely used to generate the climate prediction model. In this study, Association Rule Mining (ARM) is used to discover hidden rules in time series climate data from previous years and to analyze the relationship between the discovered rules. The dataset used in this study is a set of weather data from the Petaling Jaya observation station in Selangor for the year 2013 to 2015. This paper aims to utilize ARM for extracting behavioural patterns within the climate data that can be used to develop the prediction model for climate variability. The proposed framework is developed to provide a better approach in understanding how ARM can be used to find meaningful patterns in the climate data and generate rules that can be used to build a prediction model.