Changes for version 0.10 - 2014-01-15

  • Statistics::Data used as base so that "series testing" can be relegated to that module; so the module's own series_XXX() methods are gone; see the POD for support
  • Math::Cephes used for normal probability functions for greater precision. See p_value() description in the POD.
  • p_value() method can, if given the same args as zscore(), calculate the zscore itself and return only the p-value, rather than having to be given the z-value.
  • p_value() return value corrected for the case when zvalue = 0 and tails is optionally set to 1. Would return 1 (appropriate for the default 2-tailed setting), but now correctly returns 0.5 in this case.
  • 'pvalue' is now an alias for calling p_value()
  • values returned from zscore() and p_value() are no longer cached within the class object (was probably only useful for "series testing", which is now handled as above).
  • ccorr() method added to read or return default value used for ccorr on all tests if none is given within them (default is still 0).
  • 'ccorr' option no longer just accepts a boolean value: its actual value is implemented; it is no longer just assumed to equal 1 if defined. See OPTIONS in the POD.
  • tails() method added to read or return default value used for tails on all tests if none is given within them (default is still 2).
  • dump() method now works just like zscore() and (as now revised) p_value(): given observed, expected and variance values, will calculate the z-score and p-value itself, ahead of printing them to STDOUT. Previously it relied on the cached values from a previous call to the former methods, which is no longer useful.
  • result_string() method added: this returns what is printed by dump() but without the terminal newline character.

Modules

Data-handling and calculations for ratio of observed to standard deviation (zscore)