Time Series Analysis of Annual Rainfall, Temperature and Maize Production in Zimbabwe, (1980 To 2024).

Author

Dekeshe, William

Title

Time Series Analysis of Annual Rainfall, Temperature and Maize Production in Zimbabwe, (1980 To 2024).

Abstract

This study investigates the impact of climatic factors, especially rainfall variability and temperature fluctuations, on the yield of maize in Zimbabwe from 1980-2024. Using Time series analysis, the research explores the relationship between climate and maize yield, revealing a moderate positive correlation with rainfall (r = 0.47) and an extremely strong negative correlation with temperature (r = -0.62). To enhance prediction accuracy, the study compares the traditional ARIMA model with an advanced machine learning model, Long Short-Term Memory (LSTM) neural networks. Findings demonstrate that LSTM outperforms ARIMA, achieving a higher R-squared values (0.6793) and lower error metrics: 20.54% Mean Absolute Percentage Error (MAPE), 0.0584 Mean Absolute Error (MAE), 0.0057 Mean Squared Error (MSE), and 0.0752 Root Mean Squared Error (RMSE). The top-performing LSTM model forecasts a rise in maize production to approximately 2.3 million tonnes by 2025, followed by fluctuations, emphasizing the need for adaptive agricultural policy. The results serve to underscore the relevance of integrating climate data into models for improved risk management, policy-making, and increased food security in response to climate variability. Recommendations involve improving irrigation facilities and encouraging climate-resilient agriculture practices to counteract negative climatic effects and provide sustainable maize production in Zimbabwe.

Date

June 2025

Publisher

BUSE

Keywords

climatic factors

rainfall variability

temperature fluctuations

Supervisor

Ms J.C. Pagan’a