We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.
Start a Discussion
BI
user-pic

Should BI Offer Only Historical Reporting or Should It Also Include Predictive Analytics?

Vote 0 Votes
Brought up during a Forrester Tweetjam awhile back, should BI offer only historical or rear-view reporting or should it also include predictive analytics?

7 Replies

| Add a Reply
  • It all depends on your definition of BI.

    BI in the sense of "intelligence about your business" should absolutely include predictive analytics. The power of predictive analytics to turn uncertainty about the future into a usable probability makes it invaluable in decision making.

    BI in the sense of "Business Intelligence Software Products" should support Predictive Analytics too, at least at the level of importing PMML models from analytic environments.

    BI in the sense of a market label? No, it shouldn't. The reality is that BI means reporting and dashboards to people and we should leave it that way and use new words for new capabilities. Otherwise people will think they have the Predictive Analytics they need because they have BI and that would be a mistake.
    JT

  • In some sense, the only objective of any Business Intelligence is to help you map out tactics in the short run and strategies in the longer run, correct? In that sense all BI is for Predictive Analytics, whether you do them manually later on or the tools do them for you. Dashboards, Graphs and all BI outputs are for taking corrective action after seeing where we are and where things seem to be going.

    Of course, you can analyze past data to assign blame and get somebody fired but most of the actual USE of BI is for the future. Past Data is Information but BI's purpose is for the future, mostly!

    Of course, when we use the term Predictive Analytics we mean it in the sense of the system giving us assistance in predicting the future.

  • James is spot on... most people think BI is all about MIS reporting with flashy Dashboards! And its pretty difficult to dismiss this notion!

    I guess its time that we take MIS reporting per se off the check list for BI and have a Key Performance Indicator (KPI) driven framework for operational needs with predictive capabilities and historical reporting for strategic goals (Predictive will depend on it).

    Short answer :
    Operational+Predictive+Historical-MIS = BI(Business Indicator) :-)

  • I also agree with James' comment. I think one of the issues is that BI has the capabilities to do everything but the market place is flooded with industry related marketing terms that confuse organizations into believing that by implementing BI they will be getting everything, when in fact the entry point will probably only account for trends based analysis and reporting.

    Consequently, it becomes hard to identify where the lines cross and at what point an organization needs one, the other, or both. At the end of the day, it is important for businesses to be able to combine the types of analysis applied because the main goal of BI should be to gain general business clarity and use data to help drive decision making.

  • I've long said that the role of business intelligence is necessarily broad, if only because each business, like a puzzle, requires myriad metrics, measurements and indicators to guide its constant health and improvement. And so I'm already on record describing the utility of business intelligence tools ranging from responsive and unstructured to predictive and highly-structured. The key is knowing when to apply which. Ultimately, BI should deliver the required process-oriented, contextually-relevant, highly analytic environment around which superior decisions can be made. If predictive analytics improves those decisions, then they absolutely should be an integral part of the BI system.

    To James's point, though, using clear language to identify statistically-based, predictive models and distinguishing those from historical reports and dashboards would be very helpful. Too many today use the words "data mining" to mean everything. I'm not sure how to clean up the current mis-used nomenclature, but I support doing so.

    Brian Gentile
    Chief Executive Officer
    Jaspersoft

  • Depending on the needs of the business users and the particular subject matter, both historical and predictive analytics would be ideally suited to analyze historical trends versus future, predictive trends in an adaptive business intelligence solution. Think of it like Hurricane Tracking. On one hand you have a visual representation of where the Hurricane was in the past and based on a variety of models, you tend to predict where the Hurricane might go. In this event, you make appropriate changes to your business activities that align with the overall future trend just like cities respond to the predictive Hurricane track.
    I’ll disagree with some on this thread that claim predictive analytics is NOT part of the BI definition. Just take a look at Wikipedia and the definition of Business Intelligence, which says “BI technologies provide historical, current, and predictive views of business operations.? Predictive intelligence is coming to the forefront of the industry as advance predictive analytics become better and better over the years since the birth of business intelligence in the software industry.

  • I agree with James in how Business Intelligence is perceived today as a market value proposition. BI is typically perceived as the presentation layer to help users interact with and understand their data. The underlying engines and algorithms are typically about how that data is organized to optimize for the problem domain. In that sense, predictive analytics may be an important component of the data set being presented to the user, depending upon their use case. Similar to supply chain metrics and web analytics, the inclusion of predictive techniques should be driven by business needs. In that context, predictive analytics are just another set of data for a BI solution to take advantage of, whether it’s a credit score, propensity to purchase, or any number of predictors. Whether a vendor needs to include the algorithms directly, leverage a PMML model, or get them from a third party algorithm (from 'R' for example) is really a different discussion. A strategy of co-existence and data sharing is most likely sufficient in most cases.

    Bob Kemper, SVP Development
    PivotLink

Add a Reply

Recently Commented On

Monthly Archives

Blogs

ADVERTISEMENT