Defining a custom corpus for plotting text

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

By default, the text samples will be transformed into a vector of word counts and then modeled using Latent Dirichlet Allocation (# of topics = 50) using a model fit to a large sample of wikipedia pages. However, you can optionally pass your own text to fit the semantic model. To do this define corpus as a list of documents (strings). A topic model will be fit on the fly and the text will be plotted.

/home/docs/checkouts/readthedocs.org/user_builds/hypertools/checkouts/latest/hypertools/plot/backend.py:1353: UserWarning: Failed to switch to any interactive backend (TkAgg, QtAgg, Qt5Agg, Qt4Agg, GTK4Agg, GTK3Agg, WXAgg). Falling back to 'Agg'.
  warnings.warn(BACKEND_WARNING)

# Code source: Andrew Heusser
# License: MIT

# load hypertools
import hypertools as hyp

# load the data
text_samples = ['i like cats alot', 'cats r pretty cool', 'cats are better than dogs',
        'dogs rule the haus', 'dogs are my jam', 'dogs are a mans best friend',
        'i haz a cheezeburger?']

# plot it
hyp.plot(text_samples, 'o', corpus=text_samples)