Volume : VII, Issue : VI, June - 2017

Attribute Reduction to Enhance Classifier’s Performance: a LD Case Study

Pooja Manghirmalani Mishra, Sushil Kulkarni

Abstract :

 The objective of this study is to reduce the overlapping attributes present in a data set and to impute missing values within the same dataset. For application purpose, a dataset of Learning Disability is taken where within Learning Disabilities; there are many different types as well as a variety of tests that may be done to diagnose the problem. This study proposes a Principle Component Analysis technique for dimensionality reduction along with Artificial Neural Network’s Backpropagation algorithm to handle missing values of a dataset. This algorithm facilitates imputing the missing values in the preprocessing stage. The classification approach which is implemented by applying Winnow algorithm gives acceptable results that act as a tool for predicting the LD accurately amongst primary–grade school children. 

Keywords :

Article: Download PDF   DOI : 10.36106/ijar  

Cite This Article:

Pooja Manghirmalani–Mishra, Sushil Kulkarni, Attribute Reduction to Enhance Classifier¥s Performance: a LD Case Study, INDIAN JOURNAL OF APPLIED RESEARCH : Volume‾7 | Issue‾6 | June‾2017

Number of Downloads : 516

References :