Advances and Applications in Statistics
Volume 50, Issue 3, Pages 201 - 228
(March 2017) http://dx.doi.org/10.17654/AS050030201 |
|
ON MEASURING THE RELATIVE IMPORTANCE OF EXPLANATORY VARIABLES IN A SOFT REGRESSION METHOD
Arthur Yosef and Eli Shnaider
|
Abstract: Multiple linear regression analysis is widely used in many scientific fields to evaluate how a dependent variable is related to a set of predictors (explanatory variables). As a result, researchers and decision makers often need to assess relative importance of an explanatory variable by comparing its contribution to the contributions made by other individual explanatory variables in a regression model. Following are a number of suggested relative importance indices motivated by this definition: squared zero-order correlation, semi-partial correlation, product measure (Pratt’s measure), Johnson’s relative weight, etc.
The purpose of this article is to present additional method to compute the relative importance by soft regression. This method constitutes a tool for modeling that is more in line with human reasoning and thus is capable of serving as an effective practical mechanism for decision-making. |
Keywords and phrases: soft regression, linear regression, relative importance, fuzzy set. |
|
Number of Downloads: 399 | Number of Views: 1261 |
|