Producing decision-able information from lots of raw data often requires complex processes which must be carefully designed and architected. This presentation looks at the gap between raw data and information which helps decision-making. Most of the industry chatter about data quality focuses upon raw, granular data describing discrete events and entities inside the enterprise and around it. However, a single fact (or “cell”) of data without context is not very meaningful. Hence, we assemble many facts, aggregate them, calculate ratios and trends, often in the context of a data warehouse. It is information (data assembled and placed in proper context) which is more meaningful to the decision-maker. We will look at numerous examples between raw data and derived data (“information”) which show us trends, patterns, etc. Examples come from astronomy, military intelligence, and criminology. For example, in image analysis (in the intelligence community) we see a succession of “derived” data (sometimes facts, occasionally assumptions) as we move from pixels to clues about a country’s strategic intent. As an enterprise is inundated with more data (raw, intermediate, and advanced information) the designing and population of good metadata becomes even more important. We will look at nine distinct kinds of metadata. Not all kinds are of equal importance, but we will show the purpose and value of each. The larger the organization, and the more diverse the “institutional memory”, the more important it is to properly document the data asset for broader exploitation of its value. Ignorance of institutional memory can be disastrous. We will also look at how incorrect conclusions may be drawn because of mis-handling of the process to derive “higher level” information from the raw data.
Michael Scofield, M.B.A. is an Assistant Clinical Professor at Loma Linda University. He is a frequent speaker and author in topics of data management, data quality, data visualization, and data warehousing. He has spoken in over 27 states, Canada, Australia, and the U.K.Audiences have included 24 DAMA chapters, 5 TDWI chapters, 14 ASQ chapters (American Society for Quality), and many accounting professional organizations. He also guest lectures at several universities. His career experience includes government, manufacturing, finance, and software development. Now semi-retired, he still does pro bono data mining and data quality analysis for non-profit organizations. His greatest interest currently is data visualization, data quality assessment, and using graphic techniques to reveal business and economic behavior. Another emerging interest is data management for the “internet of things”. He also has humor published in the Los Angeles Times, and other journals.
Important to know
This meeting is included with your Buckeye DAMA membership dues.
Non-members may attend for $30, payable at the presentation hall or via PayPal at BuckeyeDama.org.