Record Details

Assessing the contribution of variables in feed forward neural network

DSpace at IIT Bombay

View Archive Info
 
 
Field Value
 
Title Assessing the contribution of variables in feed forward neural network
 
Creator PALIWAL, M
KUMAR, UA
 
Subject MULTIPLE-REGRESSION
RELATIVE IMPORTANCE
DOMINANCE ANALYSIS
COMPARING PREDICTORS
PARAMETERS
ALGORITHM
MODELS
Network weights
Prediction
Regression
Relative importance
Simulation
Multicollinearity
 
Description Neural networks are being used as tools for data analysis in a variety of applications. Neural network technique is cited in the literature as a 'Black Box' approach and criticized most for the lack of interpretability of the network weights obtained during the model building process. Some attempts have been made in the past in this direction to interpret the contributions of explanatory variables in prediction problem using the weights of neural network. In the present study, a new approach is proposed to interpret the relative importance of independent variables in neural networks and a comparison with the connection weight approach is presented. The performance of this approach is studied for various data characteristics and is found to be a better method in comparison to a well known method existing in the literature. An example from behavioral science is also considered to illustrate how the performance of the proposed approach translates to a real life situation. (C) 2011 Elsevier B.V. All rights reserved.
 
Publisher ELSEVIER SCIENCE BV
 
Date 2012-06-26T06:15:08Z
2012-06-26T06:15:08Z
2011
 
Type Article
 
Identifier APPLIED SOFT COMPUTING,11(4)3690-3696
1568-4946
http://dx.doi.org/10.1016/j.asoc.2011.01.040
http://dspace.library.iitb.ac.in/jspui/handle/100/14021
 
Language English