Visualizing high-dimensional datasets using t-SNE

Before ending this chapter, I want to introduce the reader to a very powerful algorithm called t-Distributed Stochastic Neighbor Embedding (t-SNE), which can be employed to visualize high-dimensional dataset also in 2D plots. In fact, one the hardest problems that every data scientist has to face is to understand the structure of a complex dataset without the support of graphs. This algorithm has been proposed by Van der Maaten and Hinton (in Visualizing High-Dimensional Data Using t-SNE, Van der Maaten L.J.P., Hinton G.E., Journal of Machine Learning Research 9 (Nov), 2008), and can be used to reduce the dimensionality trying to preserve the internal relationships. A complete discussion is beyond the scope of this book (but the reader can check out the aforementioned paper and Mastering Machine Learning Algorithms, Bonaccorso G., Packt Publishing, 2018), however, the key concept is to find a low-dimensional distribution so as to minimize...