UHI forecasting using machine learning algorithm

Author

Kahari, Tafadzwa

Title

UHI forecasting using machine learning algorithm

Abstract

This research aim was to develop a machine learning-based model to analyze and predict UHI effects using various environmental and urban data. Additionally, it created an interactive tool that visualizes high-risk areas and suggests appropriate mitigation strategies. The methodology used in this research integrates the use of Jupyter Notebook for exploratory data analysis and model development, Python 3.9 for implementing the machine learning algorithms, and Streamlit for deploying an interactive web interface. The development process followed the Agile software development model, which emphasizes iterative progress, continuous feedback, and adaptability to ensure the final system meets the desired objectives. Support Vector Machine (SVM) was chosen for developing the machine learning model, due to its effectiveness in high-dimensional spaces, its ability to handle both linear and non-linear data, and its robustness in classification and regression tasks. Confusion Matrix was used to evaluate the performance of the SVM model in this study. It provided a detailed breakdown of the model’s predictions against actual outcomes. Based on the evaluation metrics, particularly the confusion matrix, the SVM model achieved a high number of True Positives (TP), indicating its effectiveness in identifying instances where UHI effects were likely to occur. The model’s relatively low False Positives (FP) and False Negatives (FN) also suggest that it is capable of accurately distinguishing between areas that will experience UHI effects and those that will not. The four categories— True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN)—were used to calculate other performance metrics such as accuracy, precision, recall, and F1-score, which give further insights into the model’s overall effectiveness. The research contributes to the ongoing discourse on UHI mitigation strategies by offering a data-driven methodology that can be adapted and implemented in different urban contexts. The predictive tool developed in this study serves as a foundation for further research into improving the accuracy and applicability of machine learning models for environmental forecasting.

Date

November 2024

Publisher

BUSE

Keywords

UHI Forecasting

Machine Learning Algorithm

Supervisor

Mr. Chikwiriro