Discovering Supplier Performance Criteriafrom ERP Transactional Data withConsideration of Query Dimension

• Main Authors: Pei-Jung Wu, 吳佩蓉
• Other Authors: Gwo-Ji Sheen
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/48981928858284752910

碩士 === 國立中央大學 === 工業管理研究所 === 94 === Recently, DW (one OLAP system) provides KPIs for managers to use. DW provides multi-dimensional analysis to display the abundant information which is carried by ERP systems. But in practice, calculating from existing raw data, the indexes are definite. Under limited number of off-the-rack KPIs in DW, we want to know what more indexes we can use. Our research is inspired by demand of finding out a method to develop more KPIs. By discussing the attribute of raw data in ERP systems and formula structures in DW, we will find out the mechanism and possibility of generating KPIs automatically. In this problem, there are three sub-problems: the mechanism of classifying raw data in ERP systems; what the formula structures found in DW and the operator class; the linkage of raw data in ERP system and operator class. We will use data in one of ERP systems to develop indexes. Our cutting point is from business process point of view. To generate more indexes, we will collect key figures of InfoCubes and queries. To represent the classification of key figures and queries, we use ontology to represent them. After we classify key figures and queries, we use binary expression tree to construct formula structures by means of formulas found in InfoCubes and queries. Further, from formula structures, we try to find out operator class that can also apply to the data in ERP system. With operator class and data of ERP system, we generate possible KPI candidates. We take SAP BW for comparing base. There are thirty-seven indexes and twenty-three indexes individually for data type of quantity and currency. Our generating meaningful KPIs from ERP data for quantity and currency are forty-eight and thirty-eight. From our results of generating KPIs, we find that we exactly can generate more KPIs than indexes existing in ERP system by using our methodology.