چکیده (انگلیسی):

Data mining played vital role in comprehending, analyzing,
understanding and interpreting microarray technology expression data.
That includes search for genes that had similar or correlated patterns of
expression. For that, the feature selection was one of the frequently used
important techniques for data preprocessing. Many feature selection
algorithms had been developed. Yet the persisting problem was in
selecting optimal subset of features from the colon tumor dataset. The
use of feature selection reduced the number of features, removed
irrelevant, redundant or noise data thereby improving the accuracy,
efficiency, applicability and understandability of the learning process.
Dimensionality reduction and feature subset selection were important
components of classification techniques. In this study, the authors
presented a comparative study of existing six feature selection methods
and the proposed two algorithms of their own.