JuteBangla: A Comparative Study on Jute Yield Prediction using Supervised Machine Learning Approach based on Bangladesh Perspective

Jute provides a remarkable impact on Bangladesh’s economic growth being the second largest jute-produced country with an estimated annual production of 1.6 million tons. Estimating jute yield is required for this. Jute has been cultivated for many years, and prehistoric methods were used to calculate productivity. Bangladesh used to have a sizable market for this fiber; in 1947–1948, it held an 80% share of the export market, but by 1975–1976 it fell to only 25%. To address this problem, we gathered jute data for 15 years, from 2007 to 2021, and to forecast jute yield, we used machine learning-based classification and regression techniques. In this paper, various machine learning methods have been compared. A model is first built using the input parameters, and the estimated jute yield is then obtained. The decision tree regressor performs better than other algorithms with an average prediction accuracy of 96%, whereas other machine learning algorithms gradient boosting, k-nearest neighbor, logistic regression, and regressor have average prediction accuracy of around 92%,40%, and 42% respectively. Recently researchers have started working with Jutes and this study will help future researchers.