Comparison of Different Regression Models to Predict Mustard yield in Central Punjab

Abstract

An attempt was made to predict mustard yield over central Punjab (Ludhiana, Amritsar and Patiala) by regression models. Three statistical models have been developed for forecasting the yield of the mustard using the yield data and different weekly weather variables viz. maximum and minimum temperature, morning and evening relative humidity, sunshine hours, rainfall and number of rainy days. The sensitive period for mustard crop with respect to weather parameters were identified for different weather parameters by using correlation and selected windows with high correlation values were taken for further regression analysis. In the first Basic model, the multiple regression technique was used to predict the crop yield by using only weather parameters. The second model is Modified model, where technology trend was taken as one of the extra variable in multiple regressions. In the third model, multiple regression analysis was done using SPSS software. Regression equations were developed separately for all three models and were used to predict the mustard yield. The data for a period of (1974-2011) was used to develop the forecast model and further three years meteorological data (2012-14) was used to validate the models. For Ludhiana, among all the three models, Basic model explained up to 84 % variation in yield due to weather parameters while modified model explained highest i.e. 87% variation in mustard yield due to weather parameters. In case of Amritsar district, basic, modified and SPSS model explained 57, 59 and 62 % variation in mustard yield, respectively. In case of Patiala, the modified model was the best fit model to explain mustard yield as it explained up to 82 % variation in yield followed by SPSS (76 %) and basic model (49 %). The results reveal that modified model is best fit for Ludhiana and Patiala region, while SPSS model fits better for Amritsar region as far as mustard yield is concerned.