[−][src]Crate bhtsne
bhtsne
bhtsne
contains the implementations of the
vanilla version of t-SNE and the Barnes-Hut optimization.
Example
use bhtsne; // Dummy variables for dummy data. let n = 8; let d = 4; let theta = 0.5; let perplexity = 1.0; let max_iter = 2000; let no_dims = 2; // Load data and labels from a csv file. Labes are useful for plotting. let (mut data, labels) = match bhtsne::load_csv("data.csv", true, true, "target", 0) { (data, None) => panic!("This is not supposed to happen!"), (data, Some(labels)) => (data, labels), }; let mut y: Vec<f64> = vec![0.0; n * no_dims]; // Run t-SNE. bhtsne::run( &mut data, n, d, &mut y, no_dims, perplexity, theta, false, max_iter, 250, 250, ); // Writing the embedding to a csv file. Useful for wrappers. bhtsne::wite_csv("embedding.csv", y, 2);
Functions
load_csv | Loads data from a csv file. |
run | Performs t-SNE. |
wite_csv | Writes the embedding to a csv file. |