By Michael K. Campbell
As I mentioned in a previous tip, one of the things I’ve finally decided to look into a lot more this year is Business Intelligence and Business Performance Management. Mostly just to get a feel for what kinds of benefits these ‘disciplines’ provide, but also to help round out my SQL skills. In Deliver Actionable Insight Throughout Your Organization with Data Mining (Part 3 of 3), I gained some great insights into how to add predictive analysis into existing solutions in order to help improve business intelligence.
This webcast starts off with an overview of the differences between typical ‘retrospective’ KPIs, and then contrasts them with predictive KPIs to help show the difference between a pro-active, rather than re-active, approach to business intelligence.
An example of the difference would be a report telling a business they missed a sales goal (retrospective), versus a dashboard or report that tells a business that they’re GOING to miss a sales goal.
TIP: This webcast tells you how to use status expressions to normalize the variances between actual and anticipated KPIs to make them easier to digest by end-users and data consumers.
Then, after mentioning three likely areas to consume predictive KPIs, this webcast drops into a discussion of how to take advantage of existing functionality within SSIS in order to predict, or forecast, an outcome based on existing trends within the data.
Of course, after explaining that kind of functionality, the logical next step for this webcast was to turn to a demonstration. By showing how to create value expressions based on the percentage of customers pulled from various clusters, the first demo in this webcast showed how to contrast actual values against a goal – and then evaluate the difference in the form of indicators that can be provided as visual cues about how well a given measure is performing against a stated goal or prediction.
What was cool though, was then showing how an end-user would be able to see this within a typical scorecard (in this case, expressed in SharePoint) – as this really helped make everything that much more clear.
From this point the presenter began heavily focusing on the notion of prediction tables – something that was new to me. But the way this information was presented made a lot of sense – as it showcased how the use of prediction tables provided an easy way to add predictive, or pro-active, KPI functionality to existing business intelligence solutions. Best of all, the presenter really helped put everything into perspective by showing a typical workflow that showcased how prediction tables can be used in real-world situations.
Then, to make everything even more clear, the webcast delved into another demonstration showing how to actually execute on the data flow for this solution. The demonstration shows a sample SSIS package that utilized a Data Mining Query transform along with a conditional split to create a new prediction table that could be added to an existing Data Source View. This is then bound into an existing model as an easy-to-use resource that can be harnessed to provide predictive capabilities or KPIs in reports and dashboards.
Amazingly though, this webcast still had more to give, and provided a third demo that showcased ways to use Excel, embedded in SharePoint Reports, to provide similar predictive functionality for different users.
TIP: This webcast provides links to detailed information on how to use Excel and SharePoint to provide predictive reports. It also provides links to Data Mining Add-Ins that can be used to give Excel specialized Data Mining and predictive capabilities.
I’m not quite sure why Microsoft used a title as awkward and bulky as Deliver Actionable Insight Throughout Your Organization with Data Mining (Part 3 of 3) for this webcast, but if you’re interested in learning about ways to bridge the gap between BI and BPM, then you should definitely check this webcast out for some great options and ideas.