StatQuest: t-SNE, clearly explained!

Here’s how to create a t-SNE graph in R (this is copied from the help file for Rtsne)…

[sourcecode language="R"]
library("Rtsne")
iris_unique <- unique(iris) # Remove duplicates
iris_matrix <- as.matrix(iris_unique[,1:4])
set.seed(42) # Set a seed if you want reproducible results
tsne_out <- Rtsne(iris_matrix) # Run TSNE

# Show the objects in the 2D tsne representation
plot(tsne_out$Y,col=iris_unique$Species)
[/sourcecode]

5 thoughts on “StatQuest: t-SNE, clearly explained!

  1. I don’t suppose you could do another video on Locally Linear Embeddings, could you? I realize it’s similar to t-SNE, but it would be nice to have an intuitive yet substantive explanation of the similarities and differences.

    Great videos, btw!

    Thanks, John Strong

    • That’s a great idea. Someone else has already asked for me to do “isomap” dimension reduction, and Locally-linear embedding is related, so I’ll add it to the to-do list.

  2. GREAT VIDEOS, thank you for helping me a lot.
    But I’ve got a little question.
    Before I use the t-SNE to put datasets into graph, should I use PCA first to cluster the datasets?

Leave a Reply

Your email address will not be published. Required fields are marked *