Wednesday, June 15, 2011

Data Warehouse Planning and Selection Process

Chapter 5 of Bernhardt's book details the steps for planning a data warehouse for an organization. The purpose of this planning is to connect the various types of data – demographic, student learning, school process, and perceptions – in order to provide comprehensive data analysis (Bernhardt, p. 57). Bernhardt details seven steps necessary in the planning: 1) establishing a project team; 2) defining the scope of the data warehouse; 3) determining data readiness; 4) determining the desired data; 5) determining who is going to do the work; 6) determining the levels of access; and, 7) selecting data warehouse vendor (Bernhardt, p. 61-72). I view this process as laying the foundation for the construction of a skyscraper – there are so many unseen things going on underneath the surface which result in the successful construction of a really tall building.

Chapter 6 discusses data discovery and data mapping. At the lowest level, this process includes defining each piece of data needed in the data warehouse, its characteristics, and potential values. For example, birthdate will be entered in the format mm/dd/yyyy. The National Center for Education Statistics (NCES) refers to the entity (description), attribute (characteristics), and instance (attribute value) of the data (Bernhardt, p. 74). From a larger perspective, data discovery results in determining the reports needed from the system and cleaning up dirty data (Bernhardt, p. 73).

Chapter 7 discusses selection of analytical tools which will be used to make effective use of the data mapped through the data discovery and mapping process. Bernhardt contends that data warehouses and analytical tools must make the data accessible, consistent, and secure (Bernhardt, p. 85). The data warehouse must continuously adapt to meet users' needs. The various capabilities of a system – things such as levels of data access, intuitive use, and easy report generation – as well as vendor considerations – implementation timeframe, vendor support, and forming consortiums with like-minded districts – are also discussed (Bernhardt, p. 86-93).

Chapter 8 discusses whether a data warehouse solution should be built in-house or purchased from a vendor. Bernhardt quickly concludes that purchasing a system from a vendor is the better option, but the implementation of the system will not be successful without the support of in-house staff (Bernhardt, p. 95-96). She gives three different scenarios resulting in various levels of success or failure and emphasizes the need for IT staff and district leadership to work hand-in-hand for a successful solution.

My personal experience is consistent with Bernhardt's thoughts. There were some of my IT staff members who felt as if we could build a system ourselves and save the organization tons of money. With an extremely capable, but small development staff, this was not realistic. As a Project Manager, one of my primary, unwritten tasks was to be a liaison between the users, the management, and software vendors. The biggest asset I brought to the table was not my technical knowledge, but my ability to relate and communicate with people. This relational approach helped to smooth the waters when system implementation became challenging and ultimately resulted in a successful implementation to a new system.

Bernhardt, V.L. (2008). Translating data into information to improve teaching and learning. Larchmont, NY: Eye On Education, Inc.

No comments:

Post a Comment