Maize Yield Prediction

Overview

This repository contains the code and documentation for a project focusing on predicting maize yield for five counties in Kenya using a Simple Linear Regression Model.

Repository Structure

• code: Contains the R codes for the Simple Linear Regression Model.
• data: Includes sample datasets.
• presentation: Contains slides for non-technical audience.
• documentation: Stores project-related documents and reports.

Introduction

Background of the Study

Drought and other severe climatic conditions immensely contribute to acute food insecurity in Africa, specifically in the Horn of Africa. Over the years, an increase in the population has resulted in the destruction of the environment, especially with the clearance of forests to accommodate the growing population. Farmers have realized reduced productivity both for themselves and for the market. Unsustainable use of the available land in arid and semi-arid areas and encroachment of settlement in arable land has exacerbated the issue. The employment of technology in analyzing land use and how best to maximize productivity is key to the realization of agricultural produce throughout the region.

Problem Statement

This study addresses the challenges posed by climatic conditions and unsustainable land use. The project aims to determine the relationship between Normalized Difference Vegetation Index (NDVI) and maize yield in Kenya. NDVI is a measure of vegetation health used to determine the density of vegetation. The higher the NDVI value, the healthier the vegetation.

Objectives

General Objective

• To predict maize yield using a simple linear regression model.

Specific Objectives

• To perform exploratory data analysis to identify patterns, trends, and relationships between maize yield and NDVI.
• To fit and evaluate the model.
• To use the model for the prediction of future maize yields.

Methodology

• This study applied the theory of Simple Linear Regression to analyze the relationship between NDVI and maize yield.
• R software was used to fit the models and predict maize yield for five counties for the years 2023-2027.
• The data for the study was collected from the Ministry of Agriculture, Kilimo House, consisting of maize yield records, and from the Kenya Space Agency, consisting of NDVI measurements.

Results

A significant positive linear relationship between NDVI and maize yield was observed, indicating a significant correlation between NDVI and maize yield.

Conclusions

• An increase in NDVI causes an increase in maize yield. There is a direct relationship between NDVI and crop yield.
• Higher NDVI values correspond to higher crop yields, while lower NDVI values were associated with lower crop yields. This suggests that an increase in NDVI, as measured by satellite imagery, can be indicative of improved vegetation health and, consequently, enhanced agricultural productivity.
• Conversely, a decrease in NDVI may signify poorer vegetation conditions and lower crop yields. These results highlight the potential of utilizing NDVI data from satellite imagery as a tool to predict and monitor crop yield, offering valuable insights for agricultural management and interventions.

Recommendations

1. Explore machine learning algorithms such as Random Forest and artificial neural networks as an alternative to traditional statistical models for analyzing NDVI data and predicting crop yield.
2. Use of satellite imagery data to analyze how land topology affects vegetation health and growth in Kenya.