Page 281 -

10.1 Business Intelligence? 263

data, often referred to as “slicing and dicing”, helps to see particular patterns. By
“slicing” the OLAP cube, the analyst can zoom into a selected slice of the overall
data, e.g., only looking at sales of the iPod nano. The term “dicing” refers to the re-
ordering of dimensions, e.g., one can look at the sales of different products within a
given region or, alternatively, view the sales per region for a given product. One can
look at an OLAP cube as a dice. Using this metaphor, dicing corresponds to “throw-
ing the dice” to look at the data from another angle. The results can be viewed in
tabular form or visualized using various charts. BI products support a broad range
of charts, e.g., pie charts, bar charts, radar plots, scatter plots, speedometers, Pareto
charts, box plots, and scorecards. Any of these charts can be part of some predefined
dashboard or report.
The mainstream BI products from vendors such as IBM, Oracle, SAP, and Mi-
crosoft do not support process mining. All of the systems mentioned earlier are
data-centric and are unaware of the processes the data refers to. For example, BI
products can analyze an OLAP cube with sales data as shown in Fig. 10.1, but do
this without considering the underlying process. The sales events are immediately
aggregated without trying to distill the underlying process. BI products do not show
the end-to-end process and cannot zoom into selected parts of this process.
Another problem of mainstream BI products is that the focus is on fancy-looking
dashboards and rather simple reports, rather than a deeper analysis of the data col-
lected. This is surprising as the “I” in BI refers to “intelligence”. Unfortunately, the
business unintelligence market is dominated by large vendors that focus on monitor-
ing and reporting rather than analytics. Gartner [41] and Forrester [40] report a shift
from “measurement” to “analysis, forecasting, predictive analytics, optimization”.
However, no clear definitions of such capabilities are provided and most vendors
tend to interpret these terms in a rather deceptive manner. Data mining or statisti-
cal analysis are often added as an afterthought. For instance, IBM acquired SPSS to
add some “intelligence” to their IBM Cognos Business Intelligence offering. TIBCO
Spotfire Analytics uses the statistical programming environment called S+ (a close
relative of the well-known R open-source language for statistical computing and vi-
sualization). The open-source BI products of Jaspersoft and Pentaho can connect to
open-source data mining tools such as WEKA (Waikato Environment for Knowledge
Analysis, weka.wikispaces.com) and R (www.r-project.org).

Pentaho: An Open-Source BI Suite
To illustrate the functionality of mainstream BI products, we briefly discuss
one of the leading open-source BI products. The Pentaho BI Suite can be
downloaded from www.pentaho.com and consists of software to:
• Extract, transform, and load data from different sources (CSV files,
databases, Google Analytics, RSS, etc.)
• Design reports and dashboards
• Generate reports
• Visualize dashboards showing tabular data and a variety of charts
• View OLAP cubes in an interactive manner