Performance Analysis of Data Reduction Algorithms Using Attribute Selection in Nsl-kdd Dataset

Abstract—Data sets – both structured & unstructured, which are so large and complex that processing it using database management tools or traditional applications to derive analytical insights becomes difficult. Capturing, storage, internal search, sharing, predictive analysis and visualization are some of the major tasks being performed on such data sets. With the increa sing amount of data generated by social sharing platforms & apps, the process of data reduction has become inevitable. It involves compressing the data being generated & storing it in a data storage environment. In computer networks, these techniques have played a pivot role in increasing storage efficiency and reduced computatio nal costs. In this paper, data reduction algorithms have been applied on NSL-KDD dataset. The output of data reduction algorithm is given as an input to two classification algorithms i.e. PART and Random Forest. The aim is to find out which data reduction technique proves to be useful in enhancing the performance of the classification algorithm. Performance is compared on factors like precision, sensitivity and accuracy.Data sets – both structured & unstructured, which are so large and complex that processing it using database management tools or traditional applications to derive analytical insights becomes difficult. Capturing, storage, internal search, sharing, predictive analysis and visualization are some of the major tasks being performed on such data sets. With the increa sing amount of data generated by social sharing platforms & apps, the process of data reduction has become inevitable. It involves compressing the data being generated & storing it in a data storage environment. In computer networks, these techniques have played a pivot role in increasing storage efficiency and reduced computatio nal costs. In this paper, data reduction algorithms have been applied on NSL-KDD dataset. The output of data reduction algorithm is given as an input to two classification algorithms i.e. PART and Random Forest. The aim is to find out which data reduction technique proves to be useful in enhancing the performance of the classification algorithm. Performance is compared on factors like precision, sensitivity and accuracy.