A Guide to Principal Component Analysis (IAPT)

Author

Matthias Bartolo 0436103L

Preview:

Description of Task:

Principal component analysis (PCA) is an incredibly useful, and widely used multivariate algorithm in Machine Learning. Moreover, such algorithm is also extremely helpful in the analysis of huge datasets, whilst effectively undertaking dimensionality reduction and feature selection.

Nevertheless, such algorithm’s behaviour may not always be comprehensible, thus triggering the need for the development of a visual tool, which allow users the possibility of visualising the algorithm’s stages and data transformations, whilst offering a better understanding on the modified data. Consequently, the programmed solution also effectively portrays the PCA process as a simple convenient methodology which may be explained to students who have just completed a linear algebra or AI numerical methods course.

Additionally, the developed Jupyter notebook which outlines the aforementioned process, conveys to the students the necessary information to understand better such algorithm, whilst providing them with essential tools, to experiment and expand their knowledge. Furthermore, the notebook’s characteristics of being robust and responsive, allow students to interact with the visual plots through the plot’s minimising and maximising tools. The Jupyter notebook which incorporates the developed solution, was complemented with various famous datasets utilised by the machine learning community such as the Iris dataset, with the aim of making the students familiar with such datasets. One must note that the datasets were chosen for their distinct properties, to allow students to evaluate different experiments and infer new knowledge.

Demo - PCA Visualizations:

Video.mp4

Deliverables:

The repository includes:

1. Artefact - Directory which holds the developed Artefact.
2. Documentation - Directory which holds all materials relating to the documentation of this project.
3. Video - Directory which holds all materials relating to the creation of the video demonstration of this project

Video:

A video demonstration showcasing the developed notebook, can be accessed through one of the following links: https://youtu.be/IWYkN0AeK0o
https://drive.google.com/drive/folders/1Ms7sojfVKI_BvzDhBag98KOfX34OoKwK

Additional Info:

This project was developed as an Individual Assigned Practical Task (IAPT) for the University of Malta in 2023 supervised under Dr. Kristian Guillaumier.