Vol. 12, Issue 4 (2023)

Author(s):
Sandip Kumar, SN Singh, Ravi Ranjan Kumar, K Kumari and Subrat Keshori Behera

Abstract:
For policy maker prediction of rice yield before harvest play an important role in planning, storage, marketing, price fixation, export-import decision and distribution. However, crop yields depend on various factors like weather parameters, plant conditions, yield attributing factors and different applied inputs. This study was carried out to develop a suitable regression model for pre-harvest prediction of rice yield in the year 2019-20 in Banka district of Bihar. Different nine variables for biometrical characters and one variable as a farmers’ appraisal (crop condition) were considered for the present study. For this altogether sixty four samples were collected from farmer’s field by using multistage stratified random sampling. Out of these ten used variables, five independent variables such as X₁ (Average plant height), X₃ (Average effective number of tillers), X₆ (Applied potassium), X₇ (Irrigation level), X₉ (Average plant condition) played important role in the development of Model-5() which had minimum of coefficient of variation, RMSE, and MAE which were 11.519, 5.110, and 3.524 respectively. Adjusted R² (0.197) was observed to be the most suitable variables in the model. After model validation test, the value of percentage errors of this model had less than 12.16 and also from this average value 5.41. The average value of MAPE was close for selected model-5 which was best fitted for forecasting. Thus the estimated yield of rice in Banka district is about 41.25 q/ ha for the year 2019-20.