Data Mining for Biological and Environmental Problems

Abstract

Data Mining is a knowledge discovery from data and it treats as mining of newlineknowledge from large amount of data in every field. The algorithms are newlineimplemented using MATLAB and fuzzy logic tool box and results are evaluated newlinebased on performance parameter in both algorithms. After doing this research newlineexperiment results show that how k-means and fuzzy C means implemented on newlineprotein data set. In this research work we present the problem that show proteins are newlinehighly affiliated to each other. newlineFCM allows one piece of data to belong to two or more clusters. Results newlinebased on different clusters in both algorithms. K-means is the centroid based newlinetechnique. We are also compared k-means and FCM results in this research. newlineComparison results show that the k-means is better than FCM. With the help of this newlineresearch we can remove complexity from data sets in future. So the result shows that newlineproteins are close to each other and k-means algorithm remove data set complexity newlinewith high accuracy and less consuming time and found large sum of distance in newlineamong the statistics peak s association to FCM algorithm. Data mining techniques is newlinevery important in the analysis of real environmental data. Forest fire is important to newlinethe forest ecosystem. Only few research focus on the scientific data. It is very newlinedifficult task. In this research show that the comparison results using bagging, newlinestacking and random subspace algorithms taking place forest fire figures locate in to newlineWEKA statistics mining suite. Bagging, stacking and random subspace algorithms newlineare implemented using WEKA and experiments are behavior and consequences are newlineassessing based on performance. Finally, we compared performance of bagging, newlinestacking and random subspace algorithm. On the basis of experiments, we have newlinefound that using 20-fold cross validation then performance of the stacking with newlinedecision stump and decision table improve the prediction accuracy of classifier. So newlinestacking is better and straightforward to interpret other. Stacking algorithm built newlineaccurate classifier model a