A Survey on Dimensionality Reduction Technique
V. Arul Kumar, N. Elavarasan · 2014
Data mining is an automatic extraction of useful, often previously unknown information from large databases or data sets. The data collected from the real world applications contain lots of erroneous data. Data preprocessing is an important technique in data mining to rectify the erroneous data present in the dataset. Many data mining applications contain high dimensional data. The High dimensionality decreases the performance of the mining algorithms and increases the time and space required for processing the data. The high dimensionality issue is resolved using the Dimensionality Reduction (DR) technique. The DR is divided into two: feature selection and feature extraction. In this paper a detail survey has been carried out to know how the dimensionality problem has solved by using the two different techniques. And also various statistical measures are explained to select the most relevant features and different statistical techniques are analysed to extract the new set of features form the original features