Alpha and Beta releases do not precisely follow semantic version numbering. Backwards incompatible changes occur at secondary numeral changes (eg. 0.1 to 0.2) instead of primary numeral changes (eg. 1.0 to 2.0). Secondary numeral changes may also simply represent new features while maintaining backwards compatibility.
First adaptation from specific solution to module.
This version saw major improvements but was not released because there were some issues.
Use general formula for association ratio. The internals of Associator().find() are completely new.
The following are deprecated by this change:
- Associator().overall_ratios()
- Associator().test()
- Associator().relevant
- Associator().memo
Further encapsulate Histogram() object by providing the following:
- Histogram().reduce() (this method is also used to shorten Histogram.simplify())
- Histogram().slice()
- Histogram().nonzeros()
- Histogram().nonzero_indices
Reintroduce support for Associator().notable and Associator().significant.
Clean up excess code, removed old methods, renamed new methods to take their place.
So named because it is this version that is used for the second (and presumably final) submission for the project that initially inspired the development of this module.
- Better encapsulation and modularity.
- Better cross-platform compatibility (os.path.join()).
- Less terrible calculation of statistical significance.
- make_dir(): Create directory if non-existent.
- most_common(): Rank most common values within specified constraints.
- most_assoc(): Rank most associated pairs within specified constraints.
- extremes(): Rank most extreme associations among all data.
- savefig(): More encapsulated figure saving.
- AsciiTable(): Table methods have been split off from Analysis() into its own class.