engleski

Temporal pattern discovery in consumer behavior with REFII model

This paper represents advantages of REFII model of time series data mining in the area of consumer behavior. REF II model is based on the matrix of time series data transformation consisted from three elements: curve segment trend mark, curve segment coefficient and area beneath curve segment. REFII model is particularly suitable for combination with other data mining methods (decision trees, association rules, Bayesian networks, cluster analysis). The advantage of REFII model is illustrated on empirical data taken from textile related point-of-sale spots with the purpose of finding consumer behavior and category patterns, seasonal fluctuations and future sales projections. Complex nature of textile products and consumer influences over time, aggravate that prediction accuracy. Seasonal oscillation, day of week influence, specific month period represents simple segmentation points which pointed out useful patterns. Analysis area include sales data, statistical weather and household data in order to improve in-store supply quality together with better understanding of customer personality and specific characteristics of consumer behavior related to textile products. Crucial importance was included in time dimension. Supply and demand of textile products are very sensitive over time and dependent on many social and cultural changes. It is possible to identify related products from specific textile categories suitable for further cross-selling and market basket analysis, and to define marketing base for customer card initiation. Result received from this analysis gives significant moment for improving traditional business practice and offer different point of view on many specific occasions. Interdependence of observing textile categories and weather influence implications presents very useful decision making point for further references.

REFII; time series

nije evidentirano