1Department of Basic Sciences & Humanities, BACA, AAU, Anand, India
2College of Agricultural Information Technology, AAU, Anand, India
3Regional Research Station, AAU, Anand, India
4Department of Agricultural Statistics, BACA, AAU, Anand, India
DOI : https://doi.org/10.21276/AATCCReview.2024.12.03.366
Keywords
Abstract
An attempt to forecast paddy yield based on weather parameters, at the pre-harvest stage has been documented in this paper for middle Gujarat. Different supervised machine learning models have been tested under the study namely Linear Regression, Polynomial Regression and Neural Network. Weather variables such as maximum temperature, minimum temperature, average relative humidity, sunshine hours and accumulated rainfall have been initially used on a fortnightly basis for the period from 1980–81 to 2018–19 to train these models. Subsequent feature selection suggests that the minimum temperature and relative humidity (morning & evening) play important roles as predictors over others. Further, 37th and 39th as harvested weeks were identified with the lowest error for Linear Regression and Neural Network models respectively.