Boosting Decision Tree Algorithm for Weather Prediction

Divya Chauhan; Jawahar Thakur

doi:10.37591/joadms.v1i3.253

Abstract

Data mining is the computer assisted process which predicts behaviors and future trends by digging through and analyzing enormous sets of data and then extracting the meaningful data. It is used to answer questions that traditionally were very time consuming to resolve. This feature makes it useful to predict meteorological data that is weather prediction. Weather prediction is a necessary application in meteorology and has been one of the most scientifically and technologically challenging problems across the world in the last century. Predicting the weather is essential to help people preparing for the best and the worst of the climate. Accurately predicting the weather has been one of the most challenging problems around the world. Many weather predictions like thunderstorm prediction, rainfall prediction, predicting wind conditions, and cloud burst are major challenges for atmospheric research. The data for this work is authentic and has been collected from India Meteorological Department for city Shimla (India) from the period of January 2010 to December 2013. This paper compared three decision tree algorithms named C4.5, CART, LMT without boosting and with boosting their performance by using boosting algorithm AdaBoost. AdaBoost was run with three different iterations. It has been observed that with every increase in iteration, the accuracy of decision tree algorithm is improved. It has been also observed that LMT shows the best improvement among the decision tree algorithms after boosting. Therefore, boosted LMT is used to predict the weather of Shimla. The prediction accuracy of boosted LMT is nearly 100% when taken yearly and comparatively better than un-boosted LMT when the dataset is taken separately for each month.Keywords: data mining, decision trees, boosting, prediction, C4.5, CART, LMT,AdaBoost