Performance Evaluation of Predictive Models for Coconut Crop Production in Karnataka Using Weather Parameters
K.V. Harshith *
Department of Agricultural Statistics, Applied Mathematics and Computer Science, University of Agricultural Sciences, GKVK, Bengaluru, India.
H.R. Ragini
Department of Agricultural Statistics, Uttar Banga Krishi Viswavidyalaya, West Bengal, India.
J. Meenakshi
Department of Genetics and Plant Breeding, University of Agricultural Sciences, GKVK, Bengaluru, India.
Biradar. S. Sampat Kumar
Department of Agricultural Statistics, Applied Mathematics and Computer Science, University of Agricultural Sciences, GKVK, Bengaluru, India.
G.H. Shruthi
Department of Sericulture, University of Agricultural Sciences, GKVK, Bengaluru, India.
*Author to whom correspondence should be addressed.
Abstract
Coconut farming is a major industry in Karnataka that helps the state's agriculture sector. Karnataka is the third-largest coconut producer in India, behind Tamil Nadu and Kerala, with 3.38 billion coconuts produced on 0.42 million hectares of coconut agriculture in 2019–20, yielding an average of 8,095 nuts per hectare. These statistics come from the Directorate of Economics and Statistics. The production trends of the coconut crop in Karnataka are assessed in this study using linear, cubic, exponential, and log-logistic models. The best-fitting model is the one with the lowest Root Mean Square Error (RMSE). Between 1950 and 2019, the area under coconut crops grew cubically, while production showed loglogistic model to be the best fit, according to our data. Furthermore, we evaluate predictive models with coconut production as the dependent variable and independent factors including area, rainfall, temperature (both greatest and lowest recorded), and relative humidity. For this assessment, Stepwise MLR (SMLR) and Multiple Linear Regression (MLR) methods are used. Notably, the minimum temperature (T) and relative humidity (RH) have negative correlations with coconut production according to both stepwise regression estimates and maximum linear regression (MLR) estimates. These results imply that the minimum temperature in Karnataka and relative humidity have an inverse association with coconut crop productivity.
Keywords: Trends, Log-logistic, SMLR (Stepwise Multiple Linear Regression), Relative Humidity (RH), temperature, adverse, inverse, production