v0.6.4

Allows for single class return from a multi-class algorithm. This release will NOT be published on pypi since 0.7.0 is coming soon.

v0.6.3

Uses @jnelson16's version of textstat for Flesch scores, which uses textblob to measure sentence length.

Filters Flesch scores and average sentence lengths to be greater than -100 and less than 100 respectively (by default, filtering value can be changed with CLI argument --threshold).

v0.6.2

This hotfix allows the user to pass arguments to the ml.estimate and corpus.get_streamer functions.

v0.6.1

This hotfix catches a "division by zero" error on the sentence length analysis when no sentences are found.

v0.6.0

This version introduces readability metrics, including Flesch Reading Ease.

v0.5.0

This version significantly reorganizes and simplifies the library and framework:

• NLP analysis is moved from corpus metadata to the new nlp commands
• estimator functions have been moved to ml
• Trained estimators are now packaged as a single, .qge file
• Snakemake is now no longer a dependency; workflow management is left to the discretion of the user.

See the updated documentation for details.

Version 0.4.2: S3 Driver Bugfix

This version fixes a bug for S3 Corpus Drivers on Windows.

Bug Fix

Fixed bug related to stopwords download on import of the estimator.

Version 0.4.01: Minor update

This update fixes some problems with PyPI

Version 0.4 Release: Sentiment Analysis , Complexity, and Rounding

This version introduces a few new builtins in the corpus, including some complexity measures and sentiment analysis. It also adds in an option to round probability results from classifiers.