World Journal of Engineering Research and Technology

M.Blessa Binolin Pepsi* and R.Rohini

ABSTRACT

Classification problem is always a great challenge especially in a high dimensional data, though there are many classification problems and a feature selection (FS) algorithm has been developed in the past two decades. Feature selection algorithm results in high prediction accuracy for classification but the result is not stable when training set differs, eminently in high dimensional data. This paper proposes a new boosting based feature selection algorithm so that prediction accuracy is maintained with its stability of the selected feature subset. This is done by evaluating new Q-statistic evaluation measure. Booster in the Feature selection algorithm boosts the value of Q. Here different micro array real data sets is used to show that booster not only boost the prediction accuracy but also boost the Q –statistic. Micro array data is a collection of gene expression data. Since dealing with high dimensional data is very difficult for classification Feature Selection with boosting technique is applied for improving accuracy.

[Full Text Article] [Download Certificate]