Support of Decision Making by Business Intelligence Tools

Milena Tvrdíková Department of Applied Informatics, VŠB-Technical University Ostrava Sokolská třída 33. Ostrava 1, 701 21 Czech Republic E-mail: milena.tvrdikova@vsb.cz Abstract
Issues of business intelligence applications are
discussed. Main parts of business intelligence tools are briefly described. The attention is paid to the data warehouses and to a new tool in this field, to the operational data stores. These technologies enable to get the data of first-rate quality. Some new features of operational data stores’ applications are proposed. The role of users in the design of metadata is explained and its influence on the efficiency of information systems is emphasized. Business intelligence tools are the necessary conditions for applications of data mining, an efficient tool for getting new information from large datasets.

2. BI applications
Executive Information Systems (EIS), Customer relationship management (CRM) and Corporate Performance Management (CPM) belong among most important BI applications [15]. The applications use the multidimensional data store enabling to change quickly and flexibly individual dimensions, see Figure 1. Thus, user's view of economic reality is modified. The multidimensional data store is carried out on metadata superstructure based on the relational tables. Metadata assign lines and columns of relational databases to individual dimensions and cells in n-dimensional table. Metadata also include rules of data aggregation on individual levels of defined dimensions. This is the principle of on-line analytic data processing (OLAP) technology, see Figure 2 [13]. The OLAP technology makes the response to changes in view definitions almost immediate in most cases. The delay due to the necessary expansion of multidimensional data appears in rare situations only, e.g. in the case extensive hierarchical complexity of dimensions or in the case of increasing number of dimensions resulting in sparse data.

1. Introduction
Having sufficient amount of information has become a key factor of success in all fields of human activity. The information integrated from various sources and adjusted to certain level of detail is required for the support of decision making process [3]. Amount and structure of sources can change in time and the level of detail is various as well. It is necessary to ensure firstrate information – correspondence with reality, completeness – and to deliver it to the end user in proper time. In order to fulfill such demands, Business Intelligence (BI) solutions have been implemented in companies and institutions recently [4]. Application of BI tools was connected with data warehouse as a necessary prerequisite for the data mining. Nowadays, new technology called the operational data store is available [10]. Both the technologies enable to get the data of first-rate quality. Some new features of operational data stores applications are proposed. The role of users in the design of metadata is explained and its influence on the efficiency of information systems is emphasized.

2.1. EIS characteristics
Executive Information Systems are capable to transform considerable volumes of primary data, generated by basic company processes, into logic structures which represent the managing and decision-making processes in company [15]. Basic EIS application characteristics: multidimensional concept and data handling, obtaining of relevant data from heterogeneous data sources, tool of databases selection for OLAP and direct access to external data, advanced analytical methods, transparent organization of inner data and capability of integration with other tools, client/server architecture, support of multi-user operations (parallel access, integration and operation safety).

6th International Conference on Computer Information Systems and Industrial Management Applications (CISIM'07) 0-7695-2894-5/07 $20.00 © 2007

CRM and CPM characteristics Briefly. All the methodologies.SUPERSPACE DIMENSIONS (goods. hidden potentials and reserves. Corporate Performance Management (CPM) is based on the idea of BI tools integration into the support of company planning. 6th International Conference on Computer Information Systems and Industrial Management Applications (CISIM'07) 0-7695-2894-5/07 $20. survey features– flexible reporting. evaluate these data and modify them into new business processes influencing interaction with customer [18]. Analytic part of CRM represents the analytical processing including the data from data warehouses. According to the result. analyses) TIME Dimension 1 = period. business operations are optimized and a new strategy of the company is proposed. strategic plans and goals of a new model. cross-dimensional operations for all defined dimensions. electronic channels.1. CRM exploits the information contained in company information systems and processes it with the respect to customer characteristics. Operational part of CRM executes predefined business operations.00 © 2007 . Process activities of CPM usually involve [11]: modeling – creates a model of a company. to work with history and to anticipate future development. 2. e-mail exchange and traditional management of sales points [7]. CRM systems are usually structured into operational and analytic parts. monitoring – focuses on Key Factors of Success (Balance Scorecard). CRM effectiveness depends on the ability to gather data from different customer channels. It includes the interaction with customers through various channels: Call Centre. processing of non-standardized data and distinguishing missing and zero value. vertical axis Dimension 3 and more = viewpoints ECONOMIC INDICATORS Fig. (OLAP analysis ignores missing values regardless their source). methods and information tools needed to monitoring and managing the performance of company or an institution are parts of CPM. CRM can be considered as a way of company’s manners towards its customers. customers. (functions supplied additionally must work for any dimension). it is a strategy aimed to satisfaction of customer needs [9]. describes all connections and causal relations of company processes. Multidimensional matrix – Data cube Special features of EIS application consists of: effective processing of sparse matrices. analysis – reveals unknown trends and relations. horizontal axis Dimension 2 = variable. regions. budgeting – transforms plans into individual parts of business in operational form. planning – gives a new vision. dimension management – generic dimensionality. EIS multidimensionality enables to create new views of data placed in new relations quickly and simply to search for trend characteristics to find key indicators of deviations from planned values.2.

Personal data Accounting data Orders data SQL(OLTP) Transaction process SQL(OLTP) Transaction process SQL(OLTP) Transaction process Data Warehouse (SQL) Multidimensional (OLAP) External data Standard outputs Store balance History of supplier Customers listing Application of BI Fiscal sheets and scheduling Sale and marketing analysis Analysis of profit of company Fig. Components of a data warehouse Data warehouse forms an infrastructure for BI applications. DW sources can be diversified – data can be stored in different structures or formats. transforming (sequence of operations giving data ready to load into data warehouse). namely independent data marts (virtual DW) and integrated DW. complex solution . Such solutions are known as the data warehouse (DW). Data from DW are never erased. where data gathered through classic information systems are cumulated in individual loads. product and performance. Integrated data warehouse is centralized data storage.support of decision making. they can have a different recording philosophy or can be stored on various media [5]. Independent data marts are autonomous data stores for individual applications or their parts. see ETL. Data warehouses operate on the principle of three steps: Extracting (capability of gathering data from different sources). loading (data are put into the physical space of data warehouse. 6th International Conference on Computer Information Systems and Industrial Management Applications (CISIM'07) 0-7695-2894-5/07 $20.3. The simplest solution is to place information in one location and use it from this location for further analyses and reporting. where consistency is guaranteed by the transformation [6]. Metadata layer Source data → layer Data transformation → layer Data Reporting warehouse layer → layer Operations layer Fig.2. OLAP technology. but some data may be aggregated or deposit on external media. 3. There are two leading concepts of Data Warehousing architecture. Their disadvantages are possible inconsistency of individual stores and complicated loading processes. called also Business Intelligence reporting – provides systematic outputs for further decision making. adjusting of plans– based on monitoring and analyses. This principle of three steps is often denoted as ETL. Data warehouse A successful company needs to have an overview of basic information related to customer. Data warehouse is long-term store.00 © 2007 . where they are ready for data mining).

accessibility and accuracy. because without this context the accumulation becomes non-systematic. The following recommendations can be proposed: To store only a selected set of data derived from . however.necessity for data quality –to integrate data from various company information systems. To focus on the application of efficient tools for data transfer – The application of EII tools (Enterprise Information Integration) instead of mostly used EAI (Enterprise Application Integration) technology and seldom used ETL technology is promising in future. we must ensure four basic features of data: completeness – missing or useless data must be. Importance of metadata as a necessary description of IS content has grown considerably in recent time. providing the good description of data (attributive metadata) is great task for the users.1. way in which the content was created and the way how the content of IS is used. ODS also performs other operations in order to secure data quality and integrates relevant data from other systems at the same time. data are clustered according to individual needs. less frequently weeks or months. central store concept processing key data in real time was created and it is finding its way to commercial use. Primary reason for metadata is that they provide context to accumulation of information fragments. Data warehouse. Metadata can be divided to two main groups: technical metadata – that is information about setup of individual information systems and corresponding technical processes. identified compatibility (standardization) – all data correspond to required format. To prepare complete and detailed metadata – Individual data elements should be clearly and specifically described on operational data store level. calculating the values). Metadata make information meaningful. of operational decisions or day to day company or institution routine. users only read data. Therefore. However. when the occurrence of inconsistent. 4. integrity – all data items have fully defined relations to other data items. The shift of responsibility for data quality and their description to the users makes the information systems more effective. Data quality can be defined in several ways. integrates data from information systems. Another indispensable part of IS is a description of its content. increased productivity of executive company/institution units. Data quality Data quality is one of basic ODS characteristics.2. quicker data acquisition. attributive metadata – information about attributes of a solution. To choose the proper frequency of ODS updating – ODS is located in system transactional layer and the frequency should be a compromise between time demand of updating and permanent changes in systems.Data Warehouse features integration of data from different sources into single system. User knows the meaning and structure of data in best way. Importance of data quality grew in connection with data warehouses application. It is obvious from the architecture and the principles described above that ODS applications find their use wherever exist fragmented sources of key data which need centralized administration in order to ensure synchronizations. Centrally stored key data ensure: quality support of decision-making processes. for our purposes we will use simple intuitive definition –data of a good quality (first-rate data) correspond with reality. they are complete and consistent. that is addition of context and meaning to data items (the way of understanding in organization. To work with first-rate data. 4. Users of information systems suppose that design of metadata is the matter of supplier. recording the history of several years.00 © 2007 4. Metadata First-rate data represent essential part of any modern information system (IS). typically days. Real-time data processing is not the only goal that data stores are able to provide. such solution is insufficient. This description is called metadata (data about data). Operational data store Operational data store (ODS) was proposed to integrate selected data in real time [10]. fulfilling of regulatory demands.significance – to process data of the main importance. . as a basic BI construction unit. data are integrated in time periods. consistency – no data items contain values that represent conflicting situations. data are periodically loaded from operative information systems. . conflicting and non-compatible data from different sources is more frequent [2]. This is adequate for support of tactical or strategic decision making. data are stored at different levels of summarization. Current applications have a lack of some features that are useful and relatively easily to implement. In cases 6th International Conference on Computer Information Systems and Industrial Management Applications (CISIM'07) 0-7695-2894-5/07 $20.

(in Czech). administering. (in Czech). pp. John Wiley.. pp. IT Systems. „Magic Quadrant for CPM Suites“. [8] D. Data mining makes possible the search for relationships among the data in large datasets.Zammuto. „Building the Data Warehouse“. „EIS as necessary part of Business Intelligence“. Panec.CRMCommunity. IT System. 2005 [12] K. 2004 [5] D. (in Czech). „Teaching EIS for Students of Economics“.NET“. Gartner Research. Geldner. IT Systems. Praha. 2002. Tvrdíková.. 2002. sharing and providing first-rate data. [15] M. 6th International Conference on Computer Information Systems and Industrial Management Applications (CISIM'07) 0-7695-2894-5/07 $20. pp. pp. „Role of architecture Business Intelligence in strategic management IS/IT“. Czech Society for system integration. ”Mastering Data Mining”. In Proceedings of Informatika IV. 50-53.5. Zornes. CCB. (in Czech). BI tools bring the desired effect – transformation of data into information and information into knowledge [5]. 2006. higher effectiveness of supportive units resulting in development of new solutions.G.o. 2002. 2005 [10] Z. 2003 [9] M. (in Czech).o.... Schlegel. 2002.r. Information technologies described above increase the data quality and integrate data sources. [11] N.r. [4] R.s.r. Brno. pp. (in Czech). Linoff. Gartner Research. „Magic Quadrant for Business Intelligence Platforms. “Developing intelligent CRM”. 2006. Technical University Ostrava. DM Rewiew. Brno. 2002 [14] M.Laube. „CRM Technical Solutions and their applicability“. . Inmon. Data warehouse or operational data store are Business Intelligence tools necessary for data mining.o. Stanford Business Books. 2005 [7] M. 2000 [3] R. „Tools of Business Intelligence“. „ Magic Quadrant for CRM Customer Service and Support Applications“.r. Brno. CCB. greater believe of user into reliability of data. Gartner Research.. IT Systems. CCB. Conclusion BI tools support acquiring. The fact that data mining is capable to reveal relations previously unknown distinguish the data mining from most methods of computer data analysis.Maoz. „How to recognize our customers before competitors do it“.00 © 2007 . [2] M.304-310. IQ06“.com [17] M. pp. Geishecker. [16] http://www. (in Czech). [18] A. John Wiley & Sons. „Customer Data: Front and Center“. CCB. pp.o. Ostrava 2004. 1999. In: Proceedings of Software Development. Ericsson. Rayner. Brno. Thomsen „OLAP Solutions. Wiley Publishing. Building Multidimensional Information systems“. DM Rewiew. 2004. Tvrdíková.s. 56-58. 97-102. Ciarciello. Cenygrová.s. R.34-37. PEF MZLU Brno. L. 2006 [13] E. System Integration. [6]W. Charles River Media.s. “Business Driven Information Technology”. F. “Taking Customer Data Integration”. Klaus.24-26. Bibliography: [1] J. Berry. By incorporating BI solution to integrated information systems we achieve benefit [8]: simpler and more quality work of all units involved in decision-making process at all organizational levels. Tvrdíková.19-40. „ Building Business Intelligence Applications with . Buytendijk.