EFFECT OF VARIOUS INPUTS ON PADDY PRODUCTION - A COMPARISON OF ARTIFICIAL NEURAL NETWORKS AND LINEAR REGRESSION ANALYSIS

We analyzed the effect of chemical fertilizer, seed, biocide, farm machinery and labor hours on production of paddy (paddy rice) in the Khuzestan province in the South Western part of Iran. Here we test two methods (linear regression and neural network). We conclude that the results gotten by neural network with no hidden layer and linear regression are closed to each other. We insist that for a data set of this type the regression analysis yields more reliable results compared to a neural network. They suggest that machinery has a very clear positive effect on yield while fertilizer and labor doesn't affect on it. One can say that there is no necessity that increasing the amount of some "useful input" increase paddy production.