Comparative analysis of machine learning and time series models for cotton yield prediction in major growing districts of Karnataka

Abstract

Abstract Background Cotton is one of the most important commercial crop after food crops, especially in countries like India, where it’s grown extensively under rainfed condition. Because of its usage in multiple industries, such as textile, medicine and automobile industries, it has greater commercial importance. Cotton cultivation demands intensive management due to its explorative nature. The crop's performance is greatly influenced by prevailing weather dynamics. As climate change awareness grows, assessing how weather changes affect crop performance is essential. Crop models are a prominent tool for this purpose. Similarly, many techniques are vague and crop models are the dominant ones. Results Present study on statistical and machine learning models were compared to assess their ability to predict cotton yield across major producing districts based on long term (1990–2023) dataset on yield and weather factors. The results revealed superior performance of machine learning models such as Artificial Neural Networks (ANN) as they are iteratively trained and evaluated for higher accuracy and found that errors were within acceptable limit i.e., ± 10% and the actual and forecasted yields were in excellent agreement at both F1 and F2 stage and statistically evaluated for RMSE, nRMSE and EF, it showed good results having nRMSE value less than 10 per cent and considered as excellent for eight out of ten districts at F1 and seven districts at F2 stage because of ability of machine learning models such as ANNs to consider intricate interactive effects of weather factors. Furthermore, the tested ANN model was used to assess the importance of the dominant weather factor influencing evaluate crop performance in each district. Specifically, the use of morning relative humidity as an individual parameter and its interaction with maximum and minimum temperature had major influence on cotton yield in most of the yield predicted districts. These difference highlighted the differential interactions of weather factors in each district, which in turn affected the crop productivity. Conclusions Outcomes of the study aid in understanding the weather-related yield variability and planning crop management practices and in achieving yield sustainability under changing climatic scenarios of rainfed condition in India.