Attribute Selection in Marketing: A Rough Set Approach
Vol 22, No 1&2; Article by Sabita Mahapatra, Sreekumar & S S Mahapatra; March/June 2010
A major challenge for today's managers who operate in a technology enabled environment is to convert the large amount of data available into knowledge, and to use this knowledge to make informed and effective business decisions. Though present technologies help in marketing decisions by creating large marketing databases, most of the information may not be relevant.
Attribute reduction to eliminate superfluous or redundant data becomes an important aspect in the handling of large databases, and this calls for tools that are capable of distinguishing the various properties of the data generated.
Using an illustrative case study on the Indian cosmetic industry, this paper illustrates the advantages of the rough set approach (RSA) over conventional techniques for the extraction of decision rules from data sets, which can be useful in various marketing applications. The rule generated through the methodology can act as an 'expert', which may be referred to in future strategic decision-making. This could be done using a plug-and-play software, where the attributes are plugged in through a simulated exercise to see 'what if' scenarios in order to take business decisions. RSA involves pattern recognition through logical computational rules rather than approximation through smooth mathematical functional forms.
Statistical methods such as discriminant analysis and regression analysis make certain assumptions regarding the mathematical or Statistical methods such as discriminant analysis and regression analysis make certain assumptions regarding the mathematical or statistical properties of the data whose quality is often suspect. This paper demonstrates that almost similar accuracy can be achieved without making any mathematical or statistical assumptions regarding the data, even when the quality of data is suspect, with reliability only in the ranking of observations and not in the actual magnitudes. The findings of this study indicate that for the Indian cosmetics industry, the distribution, research and development, and miscellaneous expenditure attributes play an important role.
Reprint No 10102