Abstract
Chapter explains efficient ways of dealing with business problems of analyzing market environment and market trends under complex circumstances using heterogeneous data source. Under the assumption that used data can be expressed as time series, widely applicable multi variate model is explained together with case study in textile retail. This Chapter includes an overview of research conducted with a brief explanation of approaches and models available today. A widely applicable multi-variate decision support model is presented with advantages, limitations, and several variations for development. The explanation is based on textile retail case study with model wide range of possible applications in perspective. Complex business environment issues are simulated with explanation of several important global trends in textile retail in past seasons. Non-traditional approaches are revised as tools for a better understanding of modern market trends as well as references in relevant literature. A widely applicable multi-variate decision support model and its usage is presented through built stages and simulated. Model concept is based on specific time series transformation method in combination with Bayesian logic and Bayesian network as final business logic layer with front end interface built with open source Bayesian network tool. Explained case study provides one of the most challenging issue in textile retail: market trends seasonal/weather dependence. Separate outcomes for different scenario analysis approaches are presented on real life data from a textile retail chain located in Zagreb, Croatia. Chapter ends with a discussion about similar research’s, wide applicability of presented model with references for future research.
TopBackground
Markets are moving force of modern economy. Almost all market activities could be explained like “buying” or “selling” different merchandize and/or service. Those facts imply strong motivation of all market participants to understand, anticipate and to (if possible) manage market trends. Dealing with complexity of influences and constant market changes under complex environment leads researchers to limit research areas to smaller samples/segments or to follow simplified environments in order to achieve efficient results in pattern understanding and predictions. Although explained model is created to handle any kind of information/attribute (with assumption that data can be quantified and expressed in form of time series), proof case study is shown in one of the most complex environment, textile retail and textile supply chain.
Key Terms in this Chapter
Raise-Equal-Fall II (REFII): Time series transformation model based on data transformation to raise-equal-fall trend (name REFII represents acronym “Raise-Equal-Fall” while number “II” represents second version of proposed methodology (Klepac 2004).
AVG: Stands for average; function.
Bayesian Network: Network with nodes linked with acyclic links and trained to store knowledge in links based on Bayesian rules.
WH: Data warehouse, sometimes marked as DW or WH.
“Brick and Mortar” Store: Traditional, classic retail store (when compared to internet web shop).
Bayesian Logic: Conditional probability rule set based on so called Bayesian rule.
Directed Acyclic Graph: Directed graph with no directed cycles, it is formed by a collection of vertices and directed edges, each edge connecting one vertex to another, such that there is no way to start at some vertex and follow a sequence of edges that eventually loops back.
SNA Models: Social Network Analysis models.
Annealed MAP (AMAP): Annealed MAP algorithm, “most likely parameter configuration”.