Matplotlib 2.x By Example

t-distributed Stochastic Neighbor Embedding (t-SNE)

Since PCA is designed for datasets with a linear relationship between the features, it might not work well if a more complex, non-linear relationship exists between them. Besides, although PCA works well in separating dissimilar data points, it is not particularly competent in placing similar data points close to each other.

We are going to explore the realm of non-linear dimension reduction techniques through the use of t-SNE, which stands for t-distributed stochastic neighbor embedding. Unlike PCA, t-SNE is a probabilistic approach to dimension reduction, rather than a mathematical function that generates an exact solution. It works by building a pair-wise data similarity matrix of input ...