Growth Rate Estimation of Rabi Pulse Production of Odisha by Using Spline Regression Technique

Pulses are considered to be important crop for ensuring nutritional security in Odisha. Proper estimation of growth rate in production of pulse crops allows for more effective cropping system planning and formulation of the agricultural policy of the state. To capture any abrupt changes and the variation in data in different phases of a long time period, spline regression technique is used as it can fit different models in different segments of the time period as necessary without losing the continuity of the model. The present study deals with the estimation of growth rate of area, yield and production of all rabi pulses in Odisha by using best fit spline regression model. To fit the spline regression model, the entire period of study is divided into different segments based on the scatter plot diagram which is further confirmed by testing the significance of change in coefficient of variation between the consecutive segments by chi square test. The regression model found to be suitable from the study of scatter plot of data are linear, compound, logarithmic, power, quadratic and cubic model. The best fit model is selected on the basis of error assumption test and model fit statistics such as R2, adjusted R2 and Mean Absolute Percentage error (MAPE). The respective selected best fit model is used for the estimation of growth rates of area, yield and production of rabi pulses in Odisha for each segment and the whole period of study. Among the spline regression models, the respective linear spline regression model is found to be best fit for area, yield and production of rabi pulses and are used for growth rate estimation of these variables. It is found that though the growth rate in area and yield of rabi pulses are not significant, the growth rate of production is found to be significant for the whole period of study which shows that the interaction effect of area and yield on production seems to dominate.