Business Intelligence and Data Warehouses
Business Intelligence and Data Warehouses
The term “business intelligence” have come to define the efficiency with which an organization can improve over itself without the usual uncertainty that comes with such decisions. With the rapid expansions of many global corporations, understanding and analyzing huge amounts of transactional data (of all their activities), have become the only solution to manage their future as well as current actions. Most often multinational corporations find themselves at loss whenever rapid business decisions need to be taken which, to the desperation of many corporates, will go on to define the future of their company. And since organizational structure as well as viable business strategies will change from one geographical location to another, a single person or a board of executives can never make a foolproof plan for their company future. This is where data warehouses and business intelligence come in. And more importantly, this is where the importance of data warehouses as well as databases emerge. The most important underlying factor that we need to understand is that even though the general of purpose of these two are the same, that is to improve the business outcome, they are significantly different in their operational method as well as functions.
Databases, as used in modern businesses, are intended to serve as a data repository that stores transactional data for a few specific sets of applications. They are often systematically arranged and organized for easy data access, which is done sequentially most often. In simpler terms, databases are akin to large filing cabinets wherein company records are arranged in a particular order (such as date, alphabetical, etc) for easy access in case they are needed again. For a database to function properly data redundancy should be avoided without error. That is, for most of the business purposes, databases need to store only one instance of a particular data to be used in future transactions. Some examples of data that are usually stored in a database are, customer transaction details, employee details, health records, etc. These data are not used for analytics but are more useful for future references. And therefore, they are designed to provide easy access to information.
In contrast, data warehouses are particularly used for analytical purposes. They are primarily used in Online Analytical Processing Techniques (OLAP) since the growth of Decision Support Systems (Ma, Zhiliang, Ning, & Weihua, 2008). All business related archival data (both from past and present) are stored in a data warehouses to be used for analysis and also for developing business related decisions. For instance, if the data warehouse of an e-commerce company is well structured, the company could use the data to ascertain which product (or line of products) need to be promoted more in a particular season. Data warehouses are typically large and contains more than a single set of data. Most often, different instances of same data need to be stored in a data warehouse. For instance, in the same example, the e-commerce company might need to use the same customer transaction details (that was earlier used to predict customer behavior in a particular season) to understand the product distribution according to consumer location. So for these different purposes the same data need to be stored in such a way that the system can analyze the same data for both purposes. They are not usually as well as organized as databases.
Database Requirements for Operational and decision support data
Operational Data are usually stored in an operational data store. These databases are designed to serve as an interim location for quick reference of the mostly used or relevant data. They are structured in a way so as to answer relatively simpler demands from a small pool of structured data. These data are 1.Volatile (updated regularly or in real time) 2.Contains no archival data and are 3.Maintained as simple as possible with little summarization (Smith & Chad, 2009). Therefore, the database is required to be able to handle heavy data flow between the database and the database management system that controls it. One such proposed system uses soft database queries where the queries are flexible for easier access (Yager & Yager, 2014). These databases must be cleaned to eradicate data redundancy. Also the system must be able to discard archival data regularly. For instance, the transaction details from two weeks ago can be discarded to improve the performance of an OLTP in a business.
The structure of a data warehouse is significantly more complex. The logical structure of data warehouses entails the use of two tables 1.fact table and 2. Dimension tables. The fact table contains all the transactional details of a company while dimension table describes each entry of the fact table in detail (Chaudhuri, Dayal, & Ganti, 2001). For instance, each transaction (in fact table) will contain customer detail, transaction detail, price of the commodity, etc (which is included in dimension tables). Each of these data are stored and archived in a way so that decision support systems can understand them better.
Databases for Decision Support Systems
Some of examples where databases are used by businesses for making decisions are 1.E-commerce company trying to predict customer behavior on a particular season by tracking sale details from previous year. 2.Manager checking the transaction details of products to decide which gives them better profits in terms of number of units sold as well as profit returns from each unit. For instance when the same company manufactures two types of fridges but one sells more in a year while other is better in terms of profits, the company will have to check the database to see which product they need to produce more. 3.A company checking company databases to decide which department is consuming more resources than it needs. Also companies might check into the performance database of its employees to see who all should be removed in an event of employee downsizing.
Using data warehouses and data mining are relatively new concept. They are used by high level managers to make business decisions that are otherwise very difficult to make (Wojcik, Waldemar, & Konrad, 2011). For instance, 1.A consumer products company might mine the data warehouse of social networking sites to identify consumer interests to decide on a new product. For instance if they are looking in terms of soft drinks they could check the brands of soft drinks most consumers like to decide on the taste, look, color, and feel of their new one. 2.Insurers could decide on the premium amounts (high or low) for a particular insurance (say a house) by looking into insurance claim details from the same area. If the claims are relatively high (if it is a disaster prone area) then the insurance premiums shall be higher. 3.Bankers could use data mining to see whether a customer (or business) venturing into a different sector could repay their loans by doing what they do (and also fix the loan amount). For instance, if a IT firm is trying to set up a cloud computing server, the banker could check details of other companies to see whether the requested loan amount and projected profits are justifiable. Data warehouses and data mining are extensively used in decision support systems. For instance, they have been used to determine whether an offshore wind energy project can provide good returns by using meteorological data contained in a warehouse (Koukal, Andre, & Breitner, 2014).
References
Chaudhuri, S., Dayal, U., & Ganti, V. (2001). Database technology for decision support systems. Computer, 34(12), 48–55.
Koukal, A., Andre, K., & Breitner, M. H. (2014). Offshore Wind Energy in Emerging Countries: A Decision Support System for the Assessment of Projects. In 2014 47th Hawaii International Conference on System Sciences. http://doi.org/10.1109/hicss.2014.115
Ma, Z., Zhiliang, M., Ning, L., & Weihua, G. (2008). A decision support system for construction projects based on standardized exchanged documents. Tsinghua Science and Technology, 13(S1), 354–361.
Smith, C., & Chad, S. (2009, March). BUILDING AN OPERATIONAL DATA STORE FOR A DIRECT MARKETING APPLICATION SYSTEM. http://doi.org/10.15368/theses.2009.21
Wojcik, W., Waldemar, W., & Konrad, G. (2011). Data Mining Industrial Applications. In Knowledge-Oriented Applications in Data Mining.
Yager, R. R., & Yager, R. L. (2014). Soft Retrieval and Uncertain Databases. In 2014 47th Hawaii International Conference on System Sciences. http://doi.org/10.1109/hicss.2014.123
