Analysis and Prediction of Forest Fire occurrence and it s consequences using Data Mining techniques

Abstract

Data mining is a domain in which discovery of Knowledge is done using related data set. newlineIt involves both Supervised learning (Classification) and Unsupervised learning newline(Clustering). Data mining is a large research area to World of Environmental science, newlineMedical Science and Marketing Business. The data mining tasks requires to identify the newlinepatterns with good correlation and grouping patterns to predict the future. newline65% of world s topographical zone is forest. Each year because of forest fire 1 to 1.5% newlineof forest vegetation is under loss. In order to detect fire , several frameworks are available newlinewhich uses mainly Image Processing technology. These frame works process active fire newlineimages, in order to detect fire spots. newlineIn this work Analysis and prediction of forest fire occurrence and its consequences newlineusing data mining techniques , emphasis is given to investigate different data mining newlinetechniques that can be applied for Environmental Science. The effort is made to apply the newlineSupervised and Unsupervised algorithms to make analysis on early prediction of forest fire, newlinefire propagation and consequences of post forest fire. newlineThe work flow is as follows. newlinei)Data Collection. newlineii) Prediction of forest fire occurrence using supervised algorithm. newlineiii) Detection of fire Propagation using Unsupervised algorithms newlineiv) Prediction of Post forest fire Effects newlineThe survey is conducted on different data sets of forest fire. Based on the survey main newlineparameters used in the experimental data set are Meteorological data which consists of newlineTemperature, Humidity, Wind and Rain, Forest soil data which has Soil chemical newlinecomposition with pH values and Organic matters . The data set consists of discrete, newlinecontinuous and categorical values. Based on these parameters types, different prediction newlinemodels like Decision Trees, Rule Based Classifier are used in the research work. newlineIn Prediction task, Decision trees are considered as important predictive models. These newlineDecision tree plays a major role in order to predict the future newline