1 of 10

Comparative Analysis of Multiple Linear Regression and K-Nearest Neighbors Models in Rainfall Prediction

Salmaan Jaffer

Department of Computer Science

Supervisor: Dr. Abimbola Helen Afolayan

Co-Supervisor: Dr. Omowunmi Elizabeth Isafiade

​

​

University of the Western Cape

Click to edit presentation sub-title

2 of 10

Outline

  • Background of the Research
  • Project Objectives
  • Research Methodology
  • User Interface Design
  • Project Plan

​

3 of 10

Background of the Study

  • Climate change and its effects on Developing Nations
  • Effects of extreme rainfall on the Sustainable Development Goals (SDG)
  • Advantages of using Machine Learning Techniques

​

​

​

4 of 10

Project Objectives

The specific objectives of the study are to:

​

(i) Design a multiple linear regression and k-nearest neighbors models for predicting rainfall

(ii) Carry out a comparative analysis of the models designed in (i) using performance evaluation metrics

(iii) Implement the most accurate model in (ii) in a web-enabled environment

5 of 10

Research Methodology

Figure 1: Conceptual Framework of the WeatherWise ML.

​

6 of 10

User Interface Design

Figure 2: Proposed WeatherWise ML

​

  • Users access rainfall prediction information
  • User access reports on the model used
  • Users access past rainfall predictions

7 of 10

Functional Requirements

The WWML must

  • implement the most accurate model to predict rainfall.
  • allow users to view the daily rainfall predicted.
  • must provide a report on the models and their accuracy with the input features.
  • must provide information on the historic data the model used for prediction.

​

8 of 10

Non - Functional Requirements

  • Performance
  • Operating System
  • Reliability
  • Usability

​

9 of 10

Project Plan

10 of 10

Thank you

Presenter’s Photo Here

University of the Western Cape