Reading
The material on the following sites is important and should be read either before or after studying this section.
How to Use t-SNE Effectively
Martin Wattenberg
2016
This web article explains how t-SNE plots can be misleading, and suggests directions for properly interpreting these plots. The article provides simple examples to assist readers in learning to use t-SNE more effectively. It also underscores the importance of t-SNE hyperparameters, and how their careful selection is crucial in producing informative, insightful visualizations with this technique (Wattenberg et al., 2016).
Country Map Generated With t-SNE
This web site calls itself “An Alternative Data-Driven Country Map”, and a “t-SNE powered data exploration experiment”. It is an intuitive, interactive example of a web-based visualization system based on t-SNE. The project visualizes clusters of countries grouped by statistics for each country, including population, gross domestic product, the GINI index, and the Happy Planet Index. It is highly recommended that readers visit this web site, and experiment with the various data exploration features offered by the rich graphical user interface.
The following material is optional. However, interested readers are encouraged to peruse it.
t-SNE Clearly Explained
Kemal Erdem
Apr 13, 2020
An illustrated web article from the towardsdatascience website that explains the fundamentals of t-SNE. The article treats the basic mathematical ideas behind t-SNE. The statistical development, in particular, is somewhat involved and the reader can skip those sections, but even in those cases, the graphics help to clarify some of the more complex concepts.
Neural Network Approach Visualized With t-SNE
An application of t-SNE to neural network analysis is presented.
View: Neural Network Approach Visualized With t-SNE
Principal Component Analysis (PCA) in Python
Aditya Sharma
January 1, 2020
This web article provides a thorough, intuitive introduction to principal component analysis (PCA) without complex mathematics. The main concepts of PCA are illustrated through examples demonstrated in Python using Scikit Learn library functions and popular data sets.
Clustering with Scikit-Learn in Python
Thomas Jurczyk
September 29, 2021
This web article provides a thorough demonstration of k-means clustering on Greco-Roman authors in the ancient world. Principal component analysis is used to further analyze the results. The example in this article is illustrated with the Scikit Learn package in Python. Many code snippets are presented. Mathematical details and more advanced techniques are also provided, which the reader may skip. For the purposes of the present discussion, most benefit from the article will be drawn from the discussion of k-means clustering, principal component analysis, the explanation of the application at the intersection of literary studies and classical studies, and the instructive Python code.
The following websites may be used for reference.
PCA with Scikit Learn in Python
T-SNE with Scikit-Learn in Python
Python Code
This section uses the Python code PCA_tSNE_Example.py (Jupyter Notebook PCA_tSNE_Example.ipynb).
