Pre-Cologne Rewrite

This is essentially a rewrite of some core elements of pyzeta. Notably, the way the document-term-matrix is built has been rebuilt, providing a DTM with absolute feature counts first, then deriving DTMs with relative frequencies and binary feature counts. This is then used as the basis for calculating various versions of Zeta in one go. Visualizations remains essentially the same, with a few experimental visualizations being omitted for the time being.

Sample input and random partitions

Release 0.4.0 brings the following enhancements:

• There is now a separate, dedicated how-to document with hints on usage.
• There is now some sample input to make it easier to see what input data is needed and in what form
• A new feature has been introduced: randomly splitting the text collection into two partitions, in order to get a sense of the baseline contrasts to be expected for a given collection.

Analyses with three groups.

Analyses with three groups, based on per-text proportions of distinctive features.

First complete version after rewrite.

First complete version after rewrite.

Complete rewrite.

Complete rewrite with separate preprocessing, increased use of pandas.

Initial release.

Initial release.

Very early release.

Very early release. Added PCA based on distinctive features for three groups of texts.