Paste your data in CSV format in the Data text box below to embed it with t-SNE in two dimensions. Each row corresponds to a datapoint. You can choose to associate a label with each datapoint (it will be shown as text next to its embedding), and also a group (each group will have its own color in the embedding) (Group not yet implemented). The data can be specified either as an NxD matrix (N = number of datapoints, one per row, D = number of features), in which case a gaussian kernel will be used to compute their distances. Alternatively you can also input some distance matrix yourself.

Make sure you play with the**perplexity**, which is data specific. The perplexity is roughly speaking the number of points (note, it must be integer) that each point considers to be its neighbors while it is being embedded. High perplexities therefore enforce more global structure in the embedding, and smaller perplexities will cut up your data cloud on much finer level.

Make sure you play with the

Delimiter (default is comma (CSV)):

Learning rate: Perplexity:

Learning rate: Perplexity:

My data is: