Crop Yield Forecasting Using Discriminant Function Analysis

In this study wheat crop yield forecast models have been developed using weekly data on the weather variables such as maximum temperature, minimum temperature, rainfall and morning relative humidity. Discriminant function technique has been used for developing the forecast models. Crop yield forecast models have been developed taking the discriminant scores and trend variable as regressors and crop yield as the dependent variable. Variables (weather indices) used in the discriminant function analysis were derived through different procedures. Evaluation of the performance of the models developed using the various procedures is done by comparing the Percent Deviations of forecasts from the observed yields, Percent Standard Error (PSE), Root Mean Square Deviation (RMSE) etc. Using these criteria the model which came out to be most suitable for forecasting is based on the composite discriminant function approach.

Autorentext
I am Kaustav Aditya, a scientist in Indian Agricultural Statistics Research Institute in New Delhi, India. I have done a lot of works in the field of Agricultural and Crop Statistics and presently working in the field of Sample survey and estimation in this institute.