This is Dr. Nancy Zeliff. I will be discussing Chapter 3 with you, which is on Databases and Data Warehouses and how these support the Analytics-Driven Organization. This chapter is divided into two lectures. 3.2 is on Data Mining.
The objectives you should attain in 3.2 are: List and describe the key characteristics of a data warehouse. Define the five major types of data-mining tools. And, list key considerations in information ownership.
Operational databases in organizations have information, but that information is not organized in a way that lends itself to building business intelligence. For example, if a manager of a Victoria’s Secret wants to know “by actual versus budgeted, how many size 8 black shoes were sold last month in the Southeast and Southwest regions, compared with the same month over the last five years,” a data warehouse and data mining tools would have to be used. An operational database cannot yield these complex results. A data warehouse is a logical collection of information-gathered from many different operational databases and is used to create business intelligence that supports business analysis activities and decision-making tasks. A data warehouse is multidimensional with layers of columns and rows. Data warehouses support decision making, not transaction processing, and only support online analytical processing (OLAP). A data warehouse contains summaries of information, such as the total sales for a year by product line. A data warehouse would not contain individual transactions of customers. An organization’s operational database would process transactions, and the information contained within the operational databases is used to build summary information in a data warehouse.
Data mining tools are the software tools used to query information in a data warehouse. Data mining tools include query-and-reporting tools, artificial intelligence, multidimensional analysis tools, digital dashboards, and statistical tools.
Query and reporting tools are similar to QBE or SQL and report generators in databases. Artificial intelligence or AI tools form the basis of information discovery with fuzzy logic and neural networks. AI represents the growing convergence of various IT tools for working with information. Multidimensional analysis (MDA) tools are slice and dice techniques that allow you to view multidimensional information from different perspectives.
Digital dashboards display key information gathered from several sources on a computer screen in a format tailored to the needs and wants of an individual knowledge worker. The key items of information are called key performance indicators (KPIs) which are the most essential and important quantifiable measures used in analytics initiatives to monitor success of a business activity. Digital dashboards can operate in real time providing minute-by-minute changes. Statistical tools help apply various mathematical models to the information stored in data warehouses. Such tools help manufacturers find buying trends that help them determine which advertising strategies are working best and at what time of year.
Creating a culture of analytics in an organization doesn’t happen overnight. The Analytics Life Cycle would include gathering analytic needs from key decision makers and what type of information they need. Secondly, the information and data needed from operational databases and external databases is found.
The third step in the Analytics Life Cycle is the execution of a process known as ETL - extraction, transformation, and loading. Simply, this means when extracting needed data from its sources, transforming the data into a standardized format, and loading the transformed data into a data warehouse. The fourth step is the application of data-mining tools necessary to generate the analytics reports.
A data mart is a subset of a data warehouse in which only a focused portion of the data warehouse information is kept. Land’s End’s data warehouse was too big for employees to use. Rather, Land’s End created a data mart just for the merchandising department. Because of the smaller, more manageable data marts, knowledge workers at Land’s End made better use of available information. The same data-mining tools can be used in this smaller data mart.
Considerations in implementing a data warehouse and using data-mining tools include: Do you really need a data warehouse or does the present database environment support all the needed functions? Do all the employees need an entire data warehouse or would smaller data marts be sufficient? How up to date must the information be? And what data mining tools are needed?
Information is a resource one must manage and organize to help the organization meet its goals and objectives. These considerations are important: Strategic management support, sharing information with responsibility, and information cleanliness.
Strategic management support rests with C-level managers. The Chief Information Officer (CIO) is responsible for overseeing every aspect of an organization’s information resource. The Chief Technology Officer (CTO) oversees the underlying IT infrastructure within an organization and the user-facing technologies. The Chief Security Officer or CSO is responsible for the technical aspects of ensuring the security of information such as the development and use of firewalls, intranets, extranets, and anti-virus software. The CPO or Chief Privacy Officer ensures that information is used in an ethical way and that only the right people have access to certain types of information such as financial resources, payroll, and health care.
Other personnel in strategic management support include the data administration who plans for and oversees the development of information and monitors the information resource. The database administration is responsible for the more technical and operational aspects of managing organizational information.
Someone must accept full responsibility for providing specific pieces of information and ensuring the quality of that information.
Information cleanliness is related to ownership and who is ultimately responsible for quality and accurate information. Duplication information should be removed. The elimination of redundant information is imperative. An example of redundant information is when there are two different spellings of one’s last name.