The Impact of PCA and t-SNE on the Predictive Accuracy of k-NN, Naive Bayes, and LDA: A Study Using the Legal Medicine Legal Medicine Dataset

Abstract

Data is created in large quantities in today's digital world by many different industries, including healthcare, content creation, the internet, and business. Analysing this data to uncover relevant insights for decision-making is where machine learning (ML) algorithms come into play. However, not all features within these Legal Medicine Datasets contribute meaningfully to the construction of robust ML models. Some features may be irrelevant or have minimal impact on predictive performance. By removing these irrelevant features, the computational load on ML algorithms is alleviated. This research makes use of the open-source MNIST Legal Medicine Dataset to explore the relationship between dimensionality reduction techniques & various machine learning algorithms, such as k-Nearest Neighbours (k-NN), Naive Bayes, as well as Linear Discriminant Analysis (LDA). like t-SNE and PCA.The experimental results demonstrate the effectiveness of these ML algorithms in this context. Moreover, the study shows that incorporating PCA with ML algorithms enhances performance, especially when dealing with high-dimensional Legal Medicine Datasets.