Machine Learning Made Visual: An Educational Tool of Enhancing Machine Learning Understanding with Visualisation

Abstract

Machine learning (ML) has transformed data-driven industries, but its complexity often makes learning challenging, especially without user-friendly tools to demonstrate how ML algorithms work. Nevertheless, many existing educational resources fail to provide interactive and visual representations of ML concepts, making it difficult for users to grasp algorithmic processes and parameter impacts. To address this gap, this study introduces Machine Learning Made Visual (MLMV), an educational web tool designed to simplify and enhance understanding of ML algorithms through interactive visualisations. It focuses on two key algorithms: k-Nearest Neighbours (kNN) and k-Means Clustering; due to their simplicity and interpretability. With MLMV, users can interactively adjust algorithm parameters, such as k values in kNN and centroid initialisation in k-Means Clustering and observe their effects in real-time visual animations. The tool supports datasets with up to two features to ensure clarity in a 2-dimensional (2D) visual space, utilising technologies like Flask, NumPy, and Matplotlib for backend processing and visualisation. The results demonstrate that MLMV effectively enhances users’ comprehension of kNN and k-Means Clustering by providing clear, step-by-step visual representations of the algorithms in action. Users can explore the influence of parameter changes on algorithm performance in an intuitive and engaging manner. Ultimately, MLMV proves to be a valuable educational resource, offering a quick, responsive, and accessible platform for interactive learning without requiring significant computational resources. It fosters deeper conceptual understanding and facilitates hands-on experimentation with core ML algorithms.