berserkdisagreeableΛογισμικό & κατασκευή λογ/κού

29 Οκτ 2013 (πριν από 4 χρόνια και 8 μήνες)

99 εμφανίσεις

Business Intelligence

A Report by

Megan Amberson

Mallory Conger

Tamara Day

BA 471

May 9
, 2006

Megan Amberson

Mallory Conger

Tamara Day



Business intelligence (BI) involves technology to collect, analyze, and use data to generate relevant
information. This presentation and rep
ort talk about what business intelligence is and what really matters
to the people who use it (ease of use and reliability for example). Then we cover what the key
technologies are in business intelligence; data warehouses, data marts and data mining. Next

we cover an
insurance example in which the use of business intelligence platforms is in place to detect fraudulent
insurance claims. Next are the methods used to find patterns in data that will result in the generation of
useful information. While there a
re many methods, we focused on three. Next we cover how business
intelligence companies are evaluated with a look at Gartner’s “Magic Quadrant.” We examine what each
of Gartner’s quadrants mean, and key businesses in each quadrant. In addition we look at
how each key
player is advancing its vision into the marketplace and how well they are doing it. Finally we look at the
top ten trends for 2006 and what the future holds in store for the business intelligence market.

Introduction: What is Business Intell

The Wikipedia defines Business Intelligence (BI) as “a set of business processes for collecting and
analyzing business information. This includes the technology used in these processes, and the information
obtained from these processes (Wikipedia,

Many businesses today are finding that business intelligence is an important part of and can lead to a
sustainable competitive advantage. Some of the most common ways that BI is being put to use include,
but are not limited to retail stores usi
ng club cards to track and learn from the purchases of their
customers, airlines using frequent flyer programs to learn who their most loyal customers are and
subsequently rewarding them, or insurance companies investigating fraudulent claims. We will wal
through an example of an insurance company putting BI to work later on in the paper.

What Counts?

Various software programs have been developed to ease the work of gathering and analyzing large
quantities of data. According to Figure 1, the areas tha
t are most important when selecting a software
program to aid business intelligence ventures are ease of use, collection and analysis of real
time data,
integration with data sources, format outputs, seamless data integration and business
t. Although ease of use holds the most important spot, with approximately 65% of
respondents valuing this feature, there is not much variation between the importance of each criteria, from
about 55% of respondents for seamless integration, etc. to 65% of
respondents for ease of use.

Megan Amberson

Mallory Conger

Tamara Day


: What Counts (Weiss, 2005)

Business Intelligence Technology

Involved in the BI process are data warehouses. Wikipedia defines a data warehouse as “a repository of

integrated information from various sources, for queries and analysis (Wikipedia, 2006).” Using the
Safeway club card as an example, a data warehouse would be where a summary of all of the data from all
Safeway card customers is stored. Safeway would hav
e a database where detailed scanner data from all
Safeway card customers is stored, and the data warehouse would every so often, maybe twice a day, take
a summary of that detailed data and store it.

A subset of that data warehouse would be a data mart, wh
ich would contain specific predefined groups of
data (Wikipedia, 2006). All of the information contained in a data mart would also be contained in the
data warehouse; it would essentially be copied and isolated into its own data mart. For Safeway a data
mart might be customers that purchase Pepsi and customers that purchase Doritos.

The exploration of the data contained in data warehouses or marts in search of patterns is known as data
mining (Wikipedia, 2006). Coming back to Safeway once again, the goa
l may be to uncover a
relationship between those customers that purchase Pepsi and those customers that purchase Doritos. Do
customers that purchase Pepsi also purchase Doritos?

Business Processes Example

One example of how an insurance company would put

BI to work is shown in Figure 2. During fraud
detection process at an insurance company a claim would be subject to three different reviews. The
process would begin with the arrival of an insurance claim, which would then be reviewed by a worker. If
Megan Amberson

Mallory Conger

Tamara Day



claim were considered not suspicious then it would be sent on for processing. If it were considered
suspicious, it would be sent to the fraud detection department where it would be subject to a history
review and a comparison to similar cases. The histor
y review and comparison process is where BI
initially comes into play by pulling information from two data marts filled with previously collected data:
Policy Transactions, and Customers, to identify patterns and/or similar cases. The reviewer would use t
information from these two data marts to decide if the claim was suspicious or not. If not, then it would
be sent to the claims processing system and processed. If so, then it would be sent on for a formal
investigation. The formal investigation also
uses BI by pulling information from the firm’s entire data
warehouse: DWM insurance company, in order to answer any remaining questions and search for clues. If
the investigation determines the claim to not be fraudulent then it would be processed, otherwi
se it would
be rejected.

: Fraud Detection Business Process (Stefanov, 2005)

Finding Patterns in Data

The world today is driven by technology, resulting in volumes of data, and the need for ways to tease out

and useful information, which can then be used to make productive business decisions. Because
of this need for relevant information, data mining is needed to find patterns that can then be analyzed for
their relevance. In our research we discovered six ma
in methods or systems for finding patterns in data.
For the purposes of this paper we are only going to cover three: Neural Networks, Decision Trees and
Nonlinear Regression Models.

Neural Networks: Are “a large class of diverse systems whose architecture

to some extent imitates
structure of live neural tissue built from separate neurons (Megaputer, 2006).” In other words,
neural networks have been created to function similar to the human brain using neurons to send
Megan Amberson

Mallory Conger

Tamara Day


messages. Each layer is progressively na
rrower than the one before it, working upward until a
single neuron (message) is left. The single neuron is the result (or the prediction) of the network.
This method is very complex and requires accurate data and details to make accurate predictions.

network structure is used successfully in financial applications.

Decision Trees: Are a series of “If…, Then…” statements that ends up looking like a classification
tree. Because this method is relatively simple, it is mostly good for classification task
s in which
the results are and either one way or another. Decision trees allow business leaders to make
decisions between several choices by laying out the options and following through to the results.
(Mind Tools, 2006). They can also help business leader
s weigh risks and possible returns.

Nonlinear Regression Models: “are based on searching for a dependence of the target variable on
other variables in the form of function of some predetermined form (Megaputer, 2006).” This
method is very complex and has
a much higher statistical significance than neural networks. It
looks for specific patterns based on a given criterion. This method is highly successful in the
financial and medical fields.

Companies Involved in Business Intelligence

To get a broad pictur
e of the companies involved in business intelligence and to understand how they rate
when compared to their competitors (both direct and indirect), we presented Gartner’s Magic Quadrant
for Business Intelligence Platforms (see Figure 3). For inclusion on G
artner’s diagram companies must
meet the following four criteria:


Deliver at least 10 of 20 BI platform capabilities including: BI Infrastructure, Metadata
management, web services integration, real time data capturing, MS Office integration and
analysis with data mining.


Have a good market presence; meaning at least $10 million in annual licensing revenues.


Have significant client interest in the vendor's BI platform capabilities; measured by the number of
inquiries received from Gartner clients


Show that solutions work across the entire enterprise and go beyond single deployments.

Megan Amberson

Mallory Conger

Tamara Day


Figure 3: Magic Quadrant for BI Platforms (Schlegel, et al., 2006)

The exceptions to these requirements ar
e mainly for companies that are new to the market or that are
specialized vendors. When a vendor is accepted for evaluation it is classified into one of four categories:
Leaders, Visionaries, Niche Players, and Challengers. There are two key criteria Gartn
er uses for
evaluating vendors: Ability to execute and Completeness of Vision. We will now discuss the various
aspects of each. Ability to execute means a vendors’ ability to successfully bring its vision to market. The
components involved in a vendor’s a
bility to execute include:


Product/Service: How competitive and successful are the vendors’ products and services?


Overall Viability: How likely is the vendor to continue investing in the product and service?


Sales Execution/Pricing: Are the licensing and
maintenance options affordable (cost


Market Responsiveness and Track Record: Is the vendor able to change with the market and
customer requirements?


Market Execution: Are customers aware of vendor’s offerings?


Customer Experience: Are the vendo
r’s customers supported?


Operations: Is the company able to meet its goals and commitments?

The next criterion for evaluation is completeness of vision. Completeness of vision rates how well
vendors understand the opportunity to exploit market forces to cr
eate value for their customers and
additional opportunities for themselves.

Now that we know how vendors are evaluated, we will examine what each category is. Leaders are able to
demonstrate strong “breadth and depth of BI platform capabilities, as well
as deliver on enterprise wide
Megan Amberson

Mallory Conger

Tamara Day


implementations that support a broad BI strategy” (Gartner, 2006). In other words, they rate high on both
ability to execute and completeness of vision. Leaders are also able to offer products and services on a
global level. C
ompanies that fit into this category include Cognos and Business Objects. Visionaries have
a strong completeness of vision and are thought of as market thought
leaders and innovators, but do not
yet have sufficient scale or may have issues with their capab
ility to grow and provide consistent product
execution (their ability to execute is low). Companies that fit this category include Hyperion Solutions
and MicroStrategy. The next category is Niche Players. Niche Players are vendors who do well in a

platform, such as reporting, or that have only a limited capability to innovate. Niche vendors tend
to have less functioning in their products or may have limited implementation. In addition niche players
do not tend to have strong customer bases or a st
rong market presence. Companies that fit this category
include Actuate and Applix. The final category is Challengers. Challengers are companies that are
relatively new to the BI market and are offering products that in unison to their current businesses. T
goal is to leverage current customers since their solutions are tied to their current products and services.
Companies that fit this category include Microsoft and SAP. In the next section, we will examine key
companies and how they are involved.

at Key Companies are Involved and How?

Now we are going to look at some key companies that are involved in business intelligence and just how
they are involved. We will look at one company from each category.

: Cognos

has a strong vision and a de
dicated sales force that is able to push for quick
adoption of new products. Cognos does well executing their strategy of integration by having all their
products on a single platform. Despite this they are missing the capability of data mining. On April
9, 2006 Cognos announced that they were partnering with IBM and Google to connect their search
engines to expand its use in the business world (Gonsalves, 2006
A). This venture will allow for
easier searching of business documents and data that are stored
in Web pages or on file servers. By
making searching easier, it is hoped that more employees will use the software than just the analysts.

: Hyperion

has a good overall vision especially in linking reporting and analysis with
corporate perform
ance management (CPM) applications. Hyperion’s execution includes offering a
broad range of capabilities but software and implementation together is more expensive than other
vendors and its sharing capabilities are weaker than other platforms. In addition
, their relational
databases are under attack by leading vendors according to Kurt Schlegel (2006). One way that
Megan Amberson

Mallory Conger

Tamara Day


Hyperion is to offer a broader range of capabilities is by partnering with Microsoft to offer
complementary products. This would allow customer
s of either Microsoft or Hyperion to use the
other’s product interchangeably, according to a recent article in InformationWeek (Gonsalves, 2006

: Microsoft

has an improving vision with the release of SQL Server 2005 and its MS
Office Busi
ness Scorecard Manager. Microsoft, while late to the market, is demonstrating continued
commitment to business intelligence. It is moving in the right direction by leveraging current products
(MS Office and SQL Server), but lack partnerships that would dri
ve widespread implementation and
adoption of the business intelligence software within large corporations. The execution has gone well
with building their business intelligence platform, but adoption has been slow because they lacked a
service query c
apability until SQL Server 2005. Microsoft has the most attractive licensing and
packaging options in the market. To this point however, Microsoft has been late on delivering its
major products. Microsoft is trying to combat this lateness by incorporating
“the ability to generate an
second mirror image of an operation database” (Babcock, 2006) on their Service Pack
release. By being able to data mirror, Microsoft will be pushed into the “same high
playing field as its competitors” (Ba
bcock, 2006). Data mirroring is like having a backup copy of your
database that automatically pops up should the primary database go down. This will help boost
Microsoft’s overall rating for the next Gartner Quadrant.

Niche Player
: Actuate

has a good un
derstanding of the market but is having a hard time convincing
customers that they are more than a reporting vendor. They are also not a global player yet. Actuates’
major strategy is to advance business intelligence within the open
source community. Howev
er, open
source is not a big push for Gartner clients today. For its execution, Actuate has demonstrated that its
platform is scaleable and able to support large corporations. It also has a strong position in the
financial industry. Problems for Actuate re
sulted from being late to market and slow user adoption of
key capabilities. This has lead to limited growth.

A key issue about all the companies and how they are rated is to remember that just because a company is
in one category does not mean they are t
he right company for you. You have to determine what your
needs are and seek out a company who can best meet your needs within your budget.

Megan Amberson

Mallory Conger

Tamara Day


Top 10 Trends in 2006

While different companies choose to concentrate on specific areas of business intelligence,
there have
been ten major trends during 2006 that a study done by Knightsbridge Solutions LLC (2006) have found
to be most prevalent among business intelligence players.

Trend 1: Information Quality

Looking at a company’s information quality is the plac
e to start when solving its most important business
problems. Poor information quality can have negative consequences on regulatory compliance and
making, as well as promoting an overall sense of inefficiency within the company. To change
this, a
n information quality program can be put into place, which concentrates on measurement, quality
improvement, and verification.

Trend 2: Master Data Management

Master data management, also called MDM, is a component of an information quality program. Mas
data itself is data that describes an organization’s key business components, such as customers, products,
or vendors (Knightsbridge, 2006). This has become a popular trend in 2006 because companies realize
more and more that master data is something
that is constantly changing, so MDM requires a program
that continuously monitors, evaluates, validates, and creates master data (Knightsbridge, 2006).
Companies have to get more involved than ever in working with their master data.

Trend 3: Data Governa

Data governance provides strategic direction for a company working to improve the level of their
information quality. Data governance helps to ensure that information quality goals are achieved
(Knightsbridge, 2006). The main issue companies have wit
h data governance is in deciding who should
control their data governance efforts. Companies often use outside consultants to work on data
governance to make sure that they meet their information quality goals. An example of this would be
COBIT auditing,
which was talked about in class.

Trend 4: Enterprise Level Business Intelligence

Enterprise Level BI has not been a focused topic for many years, but more recently has become a trend for
companies who see that it is still needed. Enterprise Level BI inv
olves companies focusing on specific
areas, such as creating specific standards and performance metrics, or real
time performance management
(Knightsbridge, 2006). The issue with these types of changes is that when they occur, they need to occur
on an org
anizational level, which often takes time and money that companies either do not want to spend
or cannot afford.

Megan Amberson

Mallory Conger

Tamara Day


Trend 5: Regulatory Compliance

With Sarbanes
Oxley being such an important issue, companies have been smart about looking at their

issues and seeing how business intelligence fits into their solutions. Sarbanes
Oxley is a
major factor, but in the area of business intelligence, such compliance issues as providing certified data
and information protection are also key issues (Knightsb
ridge, 2006).

Trend 6: Enterprise Data Transparency

Enterprise data transparency means that any piece of data within an organization can be easily tracked
back to its source, similar to the concept of auditability (Knightsbridge, 2006). This has been a
trend for
companies more recently because many companies are also looking to better understand how the data has
been transformed across their organization.

Trend 7: Actionable Business Intelligence

This has been a trend of specific interest to companies
most recently because it deals with the gaps in their
measurements between the corporate and business unit levels. Companies realize that they cannot
measure something the same way on the corporate level as they do on each business level, as they are now
seeing gaps between the different measurement systems. In trying to remedy this, companies are using
what are called predictive analytics to help work with such large volumes of data between business units
(Knightsbridge, 2006).

Trend 8: Service
d Architecture

Companies see through this trend that they need to use business intelligence to help them gain a data view
of their services.

Trend 9: Rightshoring

This is one of the most important trends of 2006. Rightshoring involves getting the right m
ix of onsite,
offsite, and offshore work that delivers high quality results at a lower price while helping reduce your risk
(Knightsbridge, 2006). Similar to actionable BI, rightshoring allows companies to use analytics to
evaluate the correct sourcing op

Trend 10: Semi
structured and Unstructured Data

Companies are trying to first integrate unstructured and structured data so that the two can be analyzed
together. An example of unstructured data would be customer preferences, which would need to
integrated with structured data, and then later integrate with semi
structured data, which includes non
repeating structures and is much more difficult to bring into the picture.

Megan Amberson

Mallory Conger

Tamara Day


All of these trends in business intelligence are interrelated, and all
of them should continue on well
beyond the year 2006. Overall, there has been a consolidation of business intelligence companies, and
they are all trying to gain a major place in the market, particularly moving from strategic BI to operational
BI, which i
s also shown through these trends (Imhoff, 2006). Each company is taking advantage of the
new analytics that are available and moving quickly to remain or become a key player in the BI


Babcock, Charles. “Microsoft Pushes SQL

Server Toward Continuous Hot Availability.”

InformationWeek. (2006). Retrieved April 20, 2006

Gonsalves, Antone. “Cognos, IBM, Google Expand BI
in Enterprise.” InformationWeek. (2006).

Retrieved April 20, 2006

Gonsalves, Antone. “Microsoft, Hyperion Partner, Link Wares for BI.” Informatio
nWeek. (2006).

Retrieved April 20, 2006

Imhoff, Claudia. “Three Trends in Business Intelligence Technology: Prefect Storm or Perfect

iness Intelligence Network.

(2006). Retrieved April 24, 2006 <http://www.b

Knightsbridge Solutions LLC. “Top 10 Trend in Business Intelligence for 2006.”

(2006). Retrieved April 24, 2006

Megaputer. “Data Mining Systems.” (2006). Retrieved April 21, 2006


Mind Tools. “Decision Tree Analysis: Choosing Between Options By Projecting Likely Outcomes.”
06). Retrieved May 12, 2006 <

Schlegel, Kurt, Bill Hostmann, Andreas Bitterer, Betsy Burton. “Magic Quadrant for Business

Intelligence Platforms, 1Q06.” Gartner.

(January 9, 2006). Retrieved April 30, 2006

Stefanov, Veronica, Beate List, Birgit Korherr. “Extending UML 2 Activity Diagrams with Business

Intelligence Objects.” Vienna U
niversity of Technology. (2005). Retrieved April 25, 2006

Weiss, Peter. “Employeee Ability Shapes Business Intelligence.” InformationWeek. (August 8, 2005).
Retrieved April 20, 2006

Wikipedia Online Encyclopedia
. “Business Intelligence.” (2006). April 20, 2006