• // Business Intelligence
  • // Data Warehousing
  • // Business Analytics

Understanding (and Applying) the Styles of BI - Part 2

In Understanding and Applying the Styles of Business Intelligence – Part 1, we listed what we define as the seven styles of BI, and examined scorecards, dashboards and traditional reporting. In this post, we’ll take a look at three more styles: ad-hoc query, OLAP and advanced analytics. Again, for many of you this will be quite familiar territory. For folks newer to BI, or business users who may not have a clear sense of how BI is used, this serves as a brief primer.

Ad-hoc query


Whereas scorecards, dashboards, and reports primarily are approaches to present information to the user, ad-hoc query is a way for the user to interact with information.  It entails a user creating a custom report, data set for downloading, etc., via an information retrieval request against a data repository (which could be a set of flat files, database tables, a data warehouse, a data mart, etc.).  Ad-hoc query offers more flexibility than reporting, but requires more effort and more skill by the user.

While an ad-hoc query usually begins as a one-time endeavor, it is common for such queries to become more-than-one-time, recurring, or potentially even to morph into “production” reports.

Example analytical techniques

These basic techniques are listed here to distinguish them from advanced and predictive techniques:

  • Descriptive statistics (mean, median, mode, standard deviation, etc.)
  • Summarize data by time into week-to-date, month-to-date, etc., information
  • Compare results this year to results from the same time period last year
  • Simple time series analysis
  • Pareto analysis

Examples of ad-hoc query software tools

  • Business Objects
  • Cognos
  • Many others

OLAP (slice & dice)


OLAP is a type of ad-hoc query that emphasizes structured data exploration, typically enabled by a multi-dimensional cube.  Functionality includes drilling up and down using dimensional hierarchies, and sometimes drilling “across” from one dimension to another.  Excel pivot tables are a simple example of this type of analysis familiar to many people.

The term “OLAP,” which stands for On-Line Analytical Processing, actually refers to the multi-dimensional cube approach, not to the type of information interaction and analysis that such a structure enables.  Several new technological approaches provide the same functionality as OLAP without an actual physical implementation of a cube.  For example, in-memory analytical tools such as QlikView provide an OLAP-like experience without an actual cube.

Below is an example of a multi-dimensional cube.  This particular cube has three dimensions (month, store, and product) and only one fact (sales units).  Cubes often have many dimensions and contain many facts.  Dimensions often are hierarchical, such as a time dimension having days, months, quarters, and years.

Multi-Dimensional Cube (click to enlarge)

Example software tools

  • Microsoft SQL Server Analysis Services (SSAS) – actually, SSAS is the tool used to build the cubes
  • Oracle Hyperion Essbase – another tool used to build multi-dimensional cubes
  • QlikView
  • Microsoft PowerPivot

Advanced analytics


Advanced analytics is a way of interacting with information utilizing many different techniques used to understand the meaning of data and the root cause(s) of results. The focus is on understanding current and historical data. Some industry analytsts include a capability called “predictive modeling and data mining” defined as “enables organizations to classify categorical variables and to estimate continuous variables using advanced mathematical techniques.”  DecisionPath considers advanced analytics and predictive analytics to be separate BI styles, although some techniques and some software tools are common to both.

Example analytical techniques

  • Inferential statistics
  • Time series analysis
  • Correlation
  • Regression, both linear and non-linear
  • Collaborative filtering
  • Data mining
  • Statistical sampling, design of experiments (control groups, etc.)
  • Neural networks
  • Data visualization
  • Geographic information systems (GIS)
  • Many others

Example software tools

Typically, statistical packages like SAS and SPSS, although each advanced technique might require a specialized software product or product module optimized for that technique.  For example, SAS has a data mining product called Enterprise Miner and IBM used to have a data mining product called DB2 Intelligent Miner (which since has been absorbed into other products).  Data visualization, which is an approach to finding the meaning in data rather than a mathematical technique, is exemplified by Tableau Software.  ESRI ArcGIS is the leading product for GIS-oriented BI applications.

Up Next

We’ll finish off this series looking at predictive analytics and activity/event monitoring and alerts, and provide some advice on how and when to apply the various styles — depending on your business need.

by Bill Collins

Created by Matrix Group International, Inc. ®