Alias: null
Common Name: McNemars Test
Class: org. imirsel. m2k. evaluation. McNemarsTest

Overview: This module takes in a stream of Signal arrays and compares them with McNemar's Test.

Detailed Description: This module reads in a stream of Signal arrays with "McNemars" metadata arrays.These arrays must be of the same length in all examples and should have all been calculated on the same data set (i.e. if there are multiple folds of an experiment, the same folds of each experiment must be matched in the same order).

The "McNemars" metadata is used to perfrom the McNemars Test, which uses a Binomialdistribution and Gaussian approximation, to return the statistical probability that the each pair of algorithms has the same error function.

Ported from Prof. Stephen Cox's (sjc@cmp.uea.ac.uk) Matlab code by Kris West (kw@cmp.uea.ac.uk)

Data Handling: The input data is not modified.
Inputs:
Signal arrays
[org. imirsel. m2k. util. Signal
Signal arrays to be tested with McNemar's test (must have McNemars metadata set by an evaluator module)

Outputs:
Properties:
Verbose output
Determines whether debugging information is output to the console.

Evaluation name
The name of the evaluation being conducted (e.g. onset detection algorithms), used to crearte the name of the output file.

Output Directory
Directory to save the output file to.