Improved Distance Evaluation (IDE)

This method could be implemented for feature selection in classification problems.

This method is useful for feature selection in classification purposes. It grades features between 0 and 1 where higher value means that feature is more relevant to classification. using a cutoff value better features could be selected from the input list.
Implementation:
S=IDE2(data,classes,threshold)
[S i]=IDE2(data,classes,threshold);
[S i m]=IDE2(data,classes,threshold);

Inputs:
data: features should be listed in columns and samples should
be distributed in rows. for example a data set with 4 features
and 20 sampels should be fed as a 20x4 matrix.

classes: a one dimension matrix with length of number of
samples in data. e.g. for previous example, the classes matrix
shoud be 20x1, each cell representing a number which is related
to the class. Note: classes should be started from 1 and continuesly increase.
as an example in a case with three classes of Healthy, Fault1
and Fault2, the classes matrix should contain 1,2,3 values.

Thereshold: Threshold represents the desired cutoff value which
should be between 0 and 1. indices and selected outptus will
be generated based on this value.

Outputs:
Scores: reports the IDE scores of each feature which is bounded
between 0 and 1 and higher value means that feature is more
significant in classification. e.g, for a 20x4 input the scores
matrix will be 1x4.

Indices: based on threshold input value, returns indices of
features with score higher than threshold value.

Selected: removes features with score less than threshold value
and returns the matrix. e,g. for an input sized 20x6, and
score output = [0.2 0.3 0.9 0.7 0.6 0.3] with threshold of 0.5,
the selected matrix will be: selected=input(:,[3 4 5]);

Duration: returns the duration of calculating IDE for a given
input.