Available online www.jocpr.com Journal of Chemical and Pharmaceutical Research, 2014, 6(12):582-584 Research Article ISSN : 0975-7384 CODEN(USA) : JCPRC5 582 Temperature prediction using two-layer feed-forward backpropagation Anjaneyulu G. S. G. N. *1 , Lavanya V. 2 , Mohana Priya V. 2 and Angelin Sheeba M. 2 1 SAS, VIT University, Vellore, India 2 Department of Computer Application, SITE, VIT University, Vellore, India _____________________________________________________________________________________________ ABSTRACT Weather forecasting has become one of the challenging areas for researchers. Our day to day jobs, from agriculture to business depends on the day’s weather. Accurate weather prediction is much needed for our society today. This paper predicts daily mean temperature using two-layer feed-forward neural network and the accuracy of the predicted values is checked. Keywords: Two-layer feed-forward Neural Network, prediction. _____________________________________________________________________________________________ INTRODUCTION The frequent weather changes has made weather forecasting really important. The day to day increase in the temperature and also the heavy rains at times, draws the attention of the people towards accurate weather forecasting. In this paper, relative atmospheric humidity, wind speed, sea level pressure and rainfall are taken as the input parameters to predict temperature. The meteorological data is collected from the “Royal Netherlands Meteorological Institute (KNMI)” for Valkenburg station for a period of 2010-2013. 2. RELATED WORK: Y.Radhika and M.Shashi, 2009[1] in their paper, have predicted the atmospheric temperature. They had compared the performances of SVM (Support vector machine) and Multi Layer Perceptron (MLP). Non-linear regression was used to train the SVM and Back propagation was used to train the MLP. Their work concluded that SVM has better accuracy. Dr. S. Santhosh Baboo and I.Kadar Shereef, 2010 [2] used BPN (back propagation neural network) to predict the temperature. It was found that though neural network supported many training algorithms, BPN has highly accurate prediction than the other algorithms. Shaminder Singh, Pankaj Bhambri and Jasmeen Gill, 2011 [3] have used neural networks with genetic algorithm to predict temperature. They had integrated back propagation and genetic algorithm to train the network and followed time-series prediction. They compared the output of back propagation and back propagation with genetic algorithm and concluded that back propagation with genetic algorithm is more efficient. Parag.P.Kadu, Prof.Kishor P.Wagh and Dr.Prashant N.Chatur, 2012 [4] used ANN (Artificial Neural Network) to predict air-temperature. In formulating the ANN model, they had constructed a three-layer network for their prediction.