Crop yield prediction using linear regression: A case study on maize production

Introduction

Agriculture remains a cornerstone of global economies, especially in developing countries where it supports livelihoods and food supplies. Accurate prediction of crop yields enables better resource allocation, risk management, and policy-making. Traditional methods often rely on historical averages or expert judgment, but data-driven approaches like regression analysis offer more precision. Linear regression, a fundamental statistical technique, models the relationship between dependent variables (e.g., crop yield) and one or more independent variables (e.g., weather parameters). It assumes a linear relationship, making it simple, interpretable, and computationally efficient. This paper focuses on using linear regression for maize yield prediction, drawing from environmental data to simulate real-world applications.researchgate.netijsret.com Maize is chosen as the crop of interest due to its global significance and sensitivity to climatic factors. The study uses rainfall (in mm) and temperature (in °C) as predictors, as these are commonly correlated with yield outcomes. The objective is to demonstrate linear regression's efficacy while acknowledging its limitations in capturing non-linear effects.

Literature Review

Ali, et al. (2023) [12] thief paper author determined data mining involves the analysis of large datasets to extract meaningful information and patterns. In agriculture, data mining technologies are applied to various types of data, including agronomical, genomic, and meteorological data. This information helps in making precise and efficient decisions related to crop management practices such as irrigation, fertilization, and pest control. Machine learning, spike and slab regression analysis, and time-series analysis are some of the methods used to process agricultural data, enabling predictive analytics and better resource management Murugan R, et al. (2020) [2] numerous studies have applied regression models to crop yield prediction. For instance, a multivariate regression analysis on global crop yields using parameters like insecticides, rainfall, temperature, and yield achieved an R-squared of 0.71 with linear regression, suggesting moderate explanatory power but room for improvement with ensemble methods. Another approach employed linear regression alongside random forest models, finding that linear methods provide interpretable baselines for comparative analysis. Gupta I, (2023) [6] , Research on wheat yield prediction using hyperspectral data incorporated decision tree-based regressions, but linear models served as benchmarks. An interaction regression model introduced non-linear terms to enhance accuracy, reporting improved predictions over standard linear approaches. Clusterwise linear regression has been proposed for small farms, emphasizing localized modeling. These works underscore linear regression's role as a foundational tool, often combined with feature selection to mitigate multicollinearity. Data sources like FAOSTAT and World Bank Climate portals are frequently utilized, highlighting the importance of reliable datasets.

Methodology Data Collection and Preparation

For this study, synthetic data was generated to mimic realworld maize yield scenarios. The dataset includes 10 observations with features: rainfall (mm), temperature (°C), and yield (tons/ha). This data represents hypothetical measurements from a single growing season, with rainfall ranging from 70 to 150 mm and temperature from 21 to 31 °C. Yields vary accordingly from 35 to 75 tons/ha. In practice, data could be sourced from repositories like FAOSTAT for yield and insecticides, or World Bank for climate variables. Preprocessing would involve handling missing values, normalization, and checking for multicollinearity using Variance Inflation Factor.

Model Description Multiple linear regression models the yield (Y) as: Y=β0+β1⋅X1+β2⋅X2+ϵ

Where, • β0 is the intercept, • β1 and β2 are coefficients for rainfall (X1) and temperature (X2), • ϵ is the error term.

The model is fitted using ordinary least squares (OLS) to minimize the sum of squared residuals. Implementation was done in Python using NumPy for matrix operations.

Training and Evaluation

The design matrix X includes a column of ones for the intercept. Coefficients are solved via NumPy's lstsq function. Model performance is evaluated using R-squared, which measures the proportion of variance explained by the predictors.

Results

The fitted model yields the following coefficients:  Intercept: -25.

Conclusion

Linear regression provides a robust, straightforward method for crop yield prediction, as demonstrated by our model's strong performance. By leveraging accessible data and simple computations, it empowers farmers and policymakers.

Integrating with real-time climate data could further enhance its utility for sustainable agriculture.

References

1. Rajvanshi A. Linear regression approach to predict crop yield. Int J Adv Res Comput Sci. 2021;12(4):1-5. DOI: 10.26483/ijarcs.v12i4.6789.
2. Murugan R, Thomas FS, Geetha Shree G, Glory S, Shilpa A. Linear regression approach to predict crop yield. Asian J Comput Sci Technol. 2020;9(1):40-4. DOI: 10.51983/ajcst-2020.9.1.2152.
3. Ashwitha K, Shrinivasprasad, Spoorthi B. Advancing agricultural sustainability: Enhancing crop yield prediction through regression modeling. In: 2024 Second International Conference on Data Science and Information System (ICDSIS). Hassan: IEEE; 2024. p. 16. DOI: 10.1109/ICDSIS61070.2024.10594257.
4. Jorvekar PP, Wagh SK, Prasad JR. Advancing agricultural sustainability: Enhancing crop yield prediction through regression modeling. In: 2024 IEEE 3rd International Conference on Smart Technologies and Systems for Next Generation Computing (ICSTSN). IEEE; 2024, p. 1-6. DOI: 10.1109/ICSTSN61274.2024.10594257.
5. Fei S, Xiao S, Zhu J, Xiao Y, Ma Y. Dual sampling linear regression ensemble to predict wheat yield across growing seasons with hyperspectral sensing. Comput Electron Agric. 2023;215:108514. DOI: 10.1016/j.compag.2023.108514.
6. Gupta I. Innovations in agricultural forecasting: A multivariate regression study on global crop yield prediction. arXiv; 2023. DOI: 10.48550/arXiv.2312.02254.
7. Rajwanshi A. Crop yield prediction using leveraging linear regression. Int J Sci Res Eng Trends. 2024;10(2):289-294.
8. Li Y, Li M, Li W, Jiang L. An interaction regression model for crop yield prediction. Sci Rep. 2021;11:17823. DOI: 10.1038/s41598-021-97221-7.
9. Jorvekar PP, Wagh SK, Prasad JR. Predictive modeling of crop yields: A comparative analysis of regression techniques for agricultural yield prediction. Agric Eng Int CIGR J. 2024;26(2):1-12.
10. Figueroa MGH, Pérez MDF, Ibarra RJL, Lozada CCA. Model for predicting maize crop yield on small farms using clusterwise linear regression and GRASP. Mathematics. 2024;12(21):3356. DOI: 10.3390/math12213356.
11. Chauhan A, Singh BS, Chinmaya KA. Crop yield prediction using linear regression and random forest modelling. In: 2025 IEEE 1st International Conference on Smart and Sustainable Developments in Electrical Engineering (SSDEE). IEEE; 2025, p. 1-6.
12. Ali A, Hussain T, Tantashutikun N, Hussain N, Cocetta G. Application of smart techniques, Internet of Things and data mining for resource-use efficient and sustainable crop production. Agriculture. 2023;13(2):397. DOI: 10.3390/agriculture13020397.