Using describe to evaluate the integrity of your visualization

  run this example as a notebook — or grab the .ipynb (also linked at the bottom of the page)

The downside to using dimensionality reduction to visualize your data is that some variance will likely be removed. To help get a sense for the integrity of your low dimensional visualizations, we built the describe function. For each candidate number of dimensions, it reduces the data and correlates the pairwise Euclidean distances between observations in the reduced data with the pairwise distances in the raw (full-dimensional) data, then plots that correlation as a function of the number of dimensions.

# Code source: Andrew Heusser
# License: MIT

# import
import hypertools as hyp
import numpy as np

# load example data
data = hyp.load('weights_sample')

# plot
hyp.describe(data)