The 1990's saw the maturity of business intelligence and the Online Analytic Processing (OLAP) marketplace. OLAP...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
systems and multidimensional databases are becoming more common in application areas such as finance, supply chain, customer relationship management (CRM), and human resources. The growth of these systems has created a new challenge for the analyst or user. The problem of "I need access to data" is being replaced with "How can I make sense of all this data".
Traditional OLAP requires a user to navigate through a set of data at various levels of granularity and aggregation. Based on some level of business understanding of the data, the user picks a certain query, report, or drill down. This selection can sometimes bring about valuable insights into the data--many times, though, it does not. The user is then forced to start a new analysis based on another hypothesis regarding the data.
This hit and miss method of OLAP is referred to as passive OLAP. Basically, the system is passive in its role and serves a purpose of quick retrieval of data specified by the user query. To get users out of this hunt and peck method of analysis, the OLAP system needs to move from a passive OLAP system to an active OLAP system. The role of quick access to multidimensional data is still required. This, though, is augmented with suggested paths of analysis, based on data insights.
Data mining offers much of this insight, but removes the user too far from the analysis and decision process on what to analyze. Data mining systems take the approach of "leave the analysis to me," a more black box approach. Active OLAP uses data mining techniques and algorithms, but instead of making the decision, it suggests the analysis path to the user. It is then up to the user to take the advice, reject it fully, or accept parts of the advice.
For example, assuming a financial data mart is in place for an electronics company, and the business need is to find and analyze budget variances. An active OLAP session could analyze the multidimensional data and suggest that at the highest level, the greatest variance exists in the customer dimension between North America and the rest of the world. The user could accept this as his or her first level analysis and request the data. Based on the resulting query, the user could then drive to their next level or follow the suggestion of the active OLAP system. Let's assume that the active OLAP system now says that within the lagging North American market the greatest culprit is in the consumer electronics division. The user may choose to reject this as the next level and instead drill further down on the Geography dimension and solicit the suggestion from the active OLAP system into variances within North America.
This example can be taken to the granular level for any combination of dimensions (provided the data is available, of course). Different, though, from passive OLAP, the active OLAP system will minimize the number of dead-ends that a user may encounter. As an OLAP system moves from passive to active, the ratio of hits (queries with useful information) to misses (queries not producing useful information) increases.
Active OLAP is not for every user, but when placed in the hands of a user, they move from hunt and peck to truly analyzing the data.
Anthony Politano is CEO of the US division of MIS AG. He has over 14 years experience in the IT industry, specializing in business intelligence, data warehousing, large-scale application development and systems development methodologies. Mr. Politano was previously a Director at American Home Products, where he was responsible for building an enterprise data warehouse and multiple data marts. Prior to AHP, he was Practice Manager for data warehouse solutions with Oracle Consulting and was a Managing Consultant for James Martin & Co.