Track: Master Thesis Competition

Abstract

Maize crop yield prediction model using machine learning

Thobekile Chigwada, Mary Dzinomwa,Khulekani Sibanda, Sibonile Moyo, Belinda Ndlovu

Department of Informatics and Analytics

National University of Science and Technology

Bulawayo, Zimbabwe

N02162107w@students.nust.ac.zw,mary.dzinomwa@nust.ac.zw,khulekani.sibanda@nust.ac.zw, sibonile.moyo@nust.ac.zw, belinda.ndlovu@nust.ac.zw

Abstract

Predicting crop yield in general is critical for agricultural planning, resource allocation, and food security. The recent adverse effects of climate change on agricultural productivity have given rise to crop yield prediction which is essential in assisting farmers in anticipating the worst and preparing accordingly. Maize is one of the most popular staple foods in sub-Saharan Africa, and at the same time is the most affected by climate change. Knowing the expected maize yield in advance therefore helps those communities that are heavily dependent on maize, to prepare in advance to handle the projected situation. Despite several efforts that have been done to predict maize crop yield, the topic remains an open discussion. This paper joins the debate and uses data analytics and machine learning to build a reliable predictive model for estimating maize crop yield. Using historical maize crop yield, weather, and environmental data, the Random Forest Regressor (RFR) technique is trained to capture non-linear patterns and identify complex correlations in the data. The results of the model’s evaluation show an accuracy of 70.52% demonstrating the model’s ability to capture a substantial portion of the variability in the data. The predictive model can be useful for maize production farmers and maize grain depots, as it gives them time to plan for the future. Policymakers may benefit from the model, as it helps them to make informed policies on grain management. Future research should consider testing this model in a live environment, before packing it for deployment on a large scale.

Keywords

Random Forest Regressor, Machine Learning, Crop Yield Prediction.

Biography

Thobekile Chigwada holds a Bachelor of Science Honors degree in Computer Science. She is a Master of Science student in Big Data Science at the National University of Science and Technology, Zimbabwe. She is a distinguished educator and accomplished industry professional. With a commitment to education, her students have achieved awards in IGCSE Ordinary Level and CIE Advanced Level Cambridge examinations in Computer Science. Her research initiative centers on developing a machine learning-based maize crop yield prediction model. By bridging education and agriculture, she aims to contribute transformative insights for sustainable practices.