Main content area

A discriminant function for validation of the cluster analysis and behavioral prediction of the coffee market

Carvalho, Naiara Barbosa, Minim, Valéria Paula Rodrigues, Nascimento, Moysés, Vidigal, Márcia Cristina Teixeira Ribeiro, Ferreira, Marco Aurélio Marques, Gonçalves, Aline Cristina Arruda, Minim, Luis Antonio
Food research international 2015 v.77 pp. 400-407
cluster analysis, data collection, discriminant analysis, industrial applications, market analysis, market segmentation, marketing strategies, markets, prediction
Market segmentation is a very useful and important marketing tool for industries. However, the success of a marketing strategy designed to meet the needs of target consumer groups depends on the results of the applied methodology. Relatively little attention has been given to the reliability of the analysis used for this purpose. In this sense, the aim of the present study was to validate and predict, by means of the discriminant analysis technique, the coffee consumer groups obtained by the cluster analysis. For this, data from 210 coffee consumers obtained from market research was used. The hierarchical cluster analysis was applied to the variables related to factors that motivate the respondents to consume coffee, leading to the formation of three groups. Subsequently, the discriminant analysis technique was employed. The estimated quadratic discriminant function showed great performance with high classification accuracy (exceeding 90%) of individuals in the three different groups and a low apparent error rate (0.0476) in classification of the validation data set, confirming the existence of three distinct groups and demonstrating its ability to predict new behaviors. Thus, it appears that the discriminant analysis showed significant potential to predict the behavior of individuals, and validate and confirm results obtained by the cluster analysis, making it an alternative for industrial applications in market segmentation by marketing researchers so they can distinguish and characterize consumer groups, based on their profile and behavior, and thus create more solid and reliable strategies to meet the needs and desires of target consumer segments.