Theoretical and Empirical Analysis of ReliefF and RReliefF
Machine Learning
Evaluation of ordinal attributes at value level
Data Mining and Knowledge Discovery
| Hi-index | 0.00 |
We present a novel use of ordinal evaluation (OrdEval) algorithm as a promising technique to study various marketing phenomena. OrdEval algorithm has originated in data mining and is a general tool to analyze data with ordinal attributes, including surveys. Its many favorable features, including context sensitivity, ability to exploit meaning of ordered features and ordered response, and robustness to noise and missing values in the data, offer marketing practitioners a perspective, not available with classical analytical toolbox. We present a case study applying OrdEval algorithm on country-of-origin (COO) information. We demonstrate some interesting advantages it has to offer and show how to extract and interpret new insights allowing marketing practitioners to further optimize the management of products abroad. Data for the empirical study was gathered by means of 1225 questionnaires. Results indicate that, contrary to the classical view on COO-effects, the processing of country-related cognitions, affects and conations is a non-linear and asymmetric phenomenon. The practical implications of this finding for marketers are discussed more in detail.