Consumer involvement model of fan page: Mining from Facebook data of a real celebrity fashion brand

This paper establishes a three-layer consumer involvement model for the customers of the Facebook Fan Page of a fashion brand. A series of data mining methodology is proposed, such as data clustering, correlation coefficient and linear regression. To describe three consumer involvements, including advertisement involvement, activity involvement and product involvement, each layer designs a corresponding submodel: Ad effect submodel, engagement transfer submodel and product post submodel. Ad effect submodel uses data clustering to show the Ad cluster differences and the relationship to activities. The engagement transfer submodel fits a linear regression model between post reaches and positive engagements. Similarly, product posts can be clustered to indicate the general clusters and outliers. A data set from actual 2014 Facebook of a Taiwan celebrity fashion brand is taken as the experiment data source and the standard five KDD (Knowledge Discovery in Databases) steps are carried out. The main results of the practical data processing show: (1) the data sets of advertisement posts and product posts can be easily classified by clustering; (2) there is high correlation between post reaches and positive engagements; and (3) the general ratio (like, comment, share) of the general clusters of product posts is nearly 500 : 10 : 7.