Transcript Slide 1
Using MIS 2e
Chapter 9
Business Intelligence Systems
David Kroenke
10/23 – 6:00 AM
© Pearson Prentice Hall 2009
9-1
Study Questions
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-2
Q1 – Why do organizations need business
intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-3
Q1 – Why do organizations need business intelligence?
Computers gather and store enormous amounts of data. 403 petabytes of
new data were created in 2002. An estimated 2,500 petabytes, or 2.5
exabytes of new data were generated in 2007, 125 times the contents of all
U.S. research libraries.
All of this data is a resource which can be mined for information. Those
who do the best job of mining this information gain a competitive
advantage.
Business intelligence is information about patterns, relationships, and
trends concerning products and services related to customers,
suppliers, partners, and employees.
Business intelligence systems process, store, and provide useful
information to users who need it, when they need it.
Remember GIGO (garbage in garbage out) with regard to the effect data
quality has on business intelligence.
© Pearson Prentice Hall 2009
9-4
Q1 – Why do organizations need business intelligence?
This chart explains
the names and
amounts of computer
data measurements.
Fig 9-1 How Big is an Exabyte?
© Pearson Prentice Hall 2009
9-5
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are
available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-6
Q2 – What business intelligence systems are available?
A business intelligence (BI) system is an information system that employs
business intelligence tools to produce and deliver information which is relevant to the
user. The characteristics of a BI system depend on the BI tools in use.
BI tools are computer programs that implement a particular BI technique. The BI
techniques are categorized three ways:
Reporting tools read, process, and format data into structured reports that are delivered to
users. The processing to produce the reports includes simple operations such as filtering,
formatting, sorting, grouping, totaling, and averaging. Reporting tools are used primarily
for assessing the information content in the data.
Data-mining tools use sophisticated statistical techniques to search for patterns,
relationships, and trends among customers, suppliers, employees, and partners
concerning products and services, and then make predictions based on the results. Data
mining tools are used primarily for analyzing the information content in the data.
Knowledge-management tools store employee knowledge to make it part of the
organization’s memory and to make it available to whomever needs it. These tools are
distinguished from reporting and data mining tools because the source of the data is human
knowledge rather than recorded facts. KM tools are used primarily to access the codified
human knowledge stored in the organization’s memory.
© Pearson Prentice Hall 2009
9-7
Q2 – What business intelligence systems are available?
It’s important that you understand the difference between
business intelligence tools, applications, and systems:
A BI tool is a software program that implements the logic of a particular
procedure or process.
A BI application uses a set of BI tools on a particular type of data for a
particular purpose.
A BI system is an information system that has all five components
(hardware, software, data, procedures, and people) that delivers the
results of a BI application to users.
© Pearson Prentice Hall 2009
9-8
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-9
Q3 – What are typical reporting applications?
Reporting applications input data
from one or more sources and apply
a reporting tool to the data to
assess the information in the data.
The reporting system delivers the
results of the assessment to users.
Basic reporting (assessing)
operations include filtering,
formatting, sorting, grouping,
totaling, and averaging.
This figure shows raw data before
any reporting operations are
used.
This type of reporting can be easily
done by downloading data into
Excel, if you have less than 65,000
records and 256 fields.
Fig 9-2 Raw Sales Data
© Pearson Prentice Hall 2009
9-10
Q3 – What are typical reporting applications?
The figure on the left shows the raw sales data
sorted by customer names.
The figure on the right shows data that’s been
sorted by name and totaled by name.
Fig 9-3 Sales Data Sorted by Customer Name
© Pearson Prentice Hall 2009
Fig 9-4 Sales Data, Sorted by Customer Name & Totaled by
Customer Name
9-11
Q3 – What are typical reporting applications?
This figure shows even better information that’s been filtered and
formatted to show repeat customers.
Fig 9-5 Sales Data Filtered to Show Repeat Customers
© Pearson Prentice Hall 2009
9-12
Q3 – What are typical reporting applications?
RFM Analysis allows you to
analyze and rank customers
according to purchasing patterns
as this figure shows.
R = how recently a customer
purchased your products
F = how frequently a customer
purchases your products
M = how much money a
customer spends on your
products
The lower the score, the better
the customer.
Rank 1 to 5 for each measure
using 20% breaks.
You could rank all customers by
totaling the scores for each
measure for each customer, and
then ranking the customers based
on their overall RFM score using
pivot tables in Excel.
No model is better than a flawed
model. We do not know what we do
not know when we model. Therefore,
we are surprised by unexpected
results when using our models.
Fig 9-6 Example of RFM Score Data
© Pearson Prentice Hall 2009
9-13
Q3 – What are typical reporting applications?
Online Analytical Processing (OLAP), also called an OLAP cube, provides you
with the ability to filter, format, sort, group, total, average and perform other arithmetic
operations on a single quantitative measure across multiple categorical
dimensions. The viewer of the report can dynamically change the categorical
dimensions of the report’s format, hence the term online.
In the example below, the single quantitative measure, which is common across all
of the categorical dimensions, is Store Sales Net in dollars. The two basic
categorical dimensions are Product Family and Store Type. Product Family could
have been subdivided (e.g., types of Drink). Store Type is subdivided. Then you have
Store Type totals and Product Family totals and grand totals. This type of analysis is
easily done in Excel using pivot tables.
You could create an additional matrix to show percentage distributions to identify
the winners and losers using the 80/20 rule and conditional formatting.
Fig 9-7 OLAP Product Family by Store Type
© Pearson Prentice Hall 2009
9-14
Q3 – What are typical reporting applications?
This figure shows how you can alter the format of a report by adding
categorical dimensions to provide users with the information they need to do
their jobs. You are measuring Store Sales Net by Product Family by Store
State by Store Type, the categorical dimensions for the quantitative
measurement dimension.
This OLAP Report is using downloaded data in a pivot table in Excel.
Fig 9-8 OLAP Product Family & Store Location by Store
Type
© Pearson Prentice Hall 2009
9-15
Q3 – What are typical reporting applications?
This figure shows how you can divide data into more detail by drilling down
through the categorical dimensions, further dividing the data into more detail,
and changing the order of the categorical dimensions. You are measuring
Store Sales Net by Store Country by Store State by Store City by Product
Family by Store Type, the categorical dimensions.
Fig 9-9 OLAP Product Family & Store Location by Store Type, Drilled Down to
Show Stores in California
© Pearson Prentice Hall 2009
9-16
Q3 – What are typical reporting applications?
OLAP servers read data from an operational database, perform
some preliminary calculations, and then store the results in an OLAP
database. Normally, for performance and security reasons, the OLAP
server and the firm’s DBMS run on separate servers.
Fig 9-10 Role of OLAP Server & OLAP Database
© Pearson Prentice Hall 2009
9-17
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-18
Q4 – What are typical data-mining applications?
The purpose of data mining (i.e., information discovery in databases) is to
find patterns, relationships, and trends between categorical
dimensions using various measurement dimensions about customers,
suppliers, employees, and partners and sophisticated statistics,
mathematics, artificial intelligence and machine-learning techniques
for classification and prediction. Data reporting is performed by
managers. Data mining is performed by professional analysts.
Fig 9-11 Convergence Disciplines for Data Mining
© Pearson Prentice Hall 2009
9-19
Q4 – What are typical data-mining applications?
There are two types of data-mining techniques:
1st Type: Exploring and discovering patterns, relationships, and
trends to build a model:
No model or hypothesis (about the dimensions) exists before running the
analysis
Analysts apply data-mining techniques (to discover predictive categorical
dimensions) and then observe the results
Analysts create a hypotheses (about the categorical dimensions) after the
analysis of their measurement dimensions is completed
Cluster analysis, for example, groups categorical entities together that have
similar characteristics (i.e., what are the differences in the characteristics of
students with GPA’s => 3.0 and those with GPA’s < 3.0).
© Pearson Prentice Hall 2009
9-20
Q4 – What are typical data-mining applications?
2nd Type: Model driven data-mining:
Analysts develop a theoretical model (using categorical dimensions) prior to
their analysis.
Then they apply statistical techniques to estimate the parameters (the
measurement dimensions) of the model.
Regression analysis, for example, measures the impact of a set of variables
(causal categorical dimensions) on another variable (an effect categorical
dimension) using their measurement dimensions.
Neural networks, for example, predict values for categorical dimensions,
using their measurement dimensions, and make classifications of a primary
categorical dimension, much like cluster analysis, but use complicated nonlinear equations.
Considerable skill is required to interpret the quality, validity, reliability,
and relevancy of the model’s parameters (i.e., the quantitative values of the
measurement dimensions for the causal categorical dimensions which predict
an effect categorical dimension).
If you use too many categorical dimensions, you will over-fit the model to the
data and the model will not generalize to the population.
© Pearson Prentice Hall 2009
9-21
Q4 – What are typical data-mining applications?
Market-Basket Analysis is a data-mining tool for determining sales patterns. It helps
businesses to identify cross-selling opportunities. Three terms used with this type of
analysis are:
Support – What is the probability that two items will be purchased together? You
derive that by dividing the number of times two items occurred in the same
transaction (masks & fins) by the total number of transactions. (150/1000)=0.15
Confidence – What is the probability that A is purchased if B is purchased? You
derive that by dividing the number of times A is purchased with B (fins with masks)
by the total number of times B (masks) is purchased (150/270)=0.5556
Lift – How much more likely is A to be purchased if B is purchased? You derive
that by dividing the confidence probability (0.5556) of A (fins) by A’s base
probability (0.20): 0.5556/0.20 = 2.778.
Fig 9-12 Market-Basket Example
© Pearson Prentice Hall 2009
9-22
Q4 – What are typical data-mining applications?
A decision tree is a hierarchical arrangement of criteria that predicts a classification
value based on the criteria values. It’s an unsupervised, program controlled, datamining technique that selects the most useful attributes (categorical
dimensions) for classifying an entity (a primary categorical dimension) based
on some criterion values (measurement dimensions). The more different the
groups, the better the classification will be. The purpose is to classify the data into
groups based on the values of one or a few attributes. The program will then use
if…then rules on each of the criterion values, starting with the best predictor first, to
decide which group a record should belong to. Here are two examples.
Fig 9-13 Grades of Students from Past
MIS Class (Hypothetical Data)
© Pearson Prentice Hall 2009
Fig 9-14 Credit Score Decision Tree
9-23
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses
and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-24
Q5 – What is the purpose of data warehouses and data marts?
Firms extract operational data into data warehouses to resolve missing
data values and inconsistent data, and to standardize data formats
within operational data and between operational data and data purchased
from third-party vendors to prepare, store, and manage data specifically for
data mining and analyses. Then the “cleaned data” can be extracted into
data marts.
The data warehouse and data mart DBMS, database, and metadata are
distinct from the operational DBMS and database.
Fig 9-15 Components of a Data Warehouse
© Pearson Prentice Hall 2009
9-25
Q5 – What is the purpose of data warehouses and data marts?
Clickstream data refers to the
recorded clicking behavior of
customers on Web sites, which
can produce too much data.
The curse of dimensionality is
over-fitting the model to the
sample data by using too many
categorical dimensions, which
makes the model useless as a
predictor using different data.
© Pearson Prentice Hall 2009
The figure on the left, lists some of the
data that’s readily available for
purchase from data vendors
Some of the problems companies
experience with operational data are
shown in the figure below.
Remember GIGO regardless of the
quality of the analysts and the analyses.
Be careful. Analysts can get impressive,
convincing results with incorrect
analyses, flawed data, and
idiosyncratic models.
9-26
Q5 – What is the purpose of data warehouses and data marts?
Here’s the difference between a data warehouse and a data mart:
A data warehouse extracts operational data and combines the extract
with purchased data. It cleans and processes the data as necessary and
then makes the data available to data marts.
A data mart extracts a segment of the cleaned, processed data from the
data warehouse and focuses on a particular analysis or functional area
of an organization.
Fig 9-18 Data Mart Examples
© Pearson Prentice Hall 2009
9-27
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management
applications?
Q7 – How are business intelligence applications delivered?
© Pearson Prentice Hall 2009
9-28
Q6 – What are typical knowledge-management applications?
Knowledge management (KM) is the process of creating a knowledgebase by codifying employee knowledge and organizing and indexing this
knowledge in organizational documents and making that knowledge-base
available to all employees.
Reporting and data mining create new information. KM systems share
existing codified human knowledge and preserve organizational memory.
Collaboration systems are concerned with document creation and change
management. KM systems are concerned with maximizing the use of
codified human organizational memory.
The three major categories of knowledge assets are data, documents,
and employees.
© Pearson Prentice Hall 2009
9-29
Q6 – What are typical knowledge-management applications?
Two key technologies for sharing content in KM systems are:
Indexing - the single most important content function in KM applications. It’s
an easily accessible and robust means of determining if content exists and
includes a link to obtain the content. Users need keyword search providing
quick response and high document relevancy. The organization has to provide
KM systems for internal documentation because commercial internet search
engines (e.g., Google) cannot index documents protected by an organization’s
firewall.
RSS, Real Simple Syndication - a standard for subscribing to content sources
on Web sites. It is an email system for content. RSS uses an RSS Reader
program which helps users as follows:
You subscribe to an RSS system for specific content sources.
The RSS reader periodically checks these sources for new or updated
content.
The RSS reader uses an RSS feed process to place content summaries in
your RSS inbox with a link to the full content.
You process your RSS inbox just like your email inbox.
© Pearson Prentice Hall 2009
9-30
Q6 – What are typical knowledge-management applications?
This figure shows a typical RSS reader. The left pane shows RSS sources
which provide an RSS feed to your RSS inbox. Entries are grouped into
categories predetermined by the user.
Fig 9-19 Interface of a Typical RSS Reader
© Pearson Prentice Hall 2009
9-31
Q6 – What are typical knowledge-management applications?
Blogs provide an easy way to share knowledge as seen in this figure. You
can use RSS feeds to subscribe to thousands of blogs.
Fig 9-20 Blog Posts of SharePoint Team Member
© Pearson Prentice Hall 2009
9-32
Q6 – What are typical knowledge-management applications?
Another form of knowledge management is an expert system. Here are
characteristics about expert systems along with some of their problems.
Characteristics
They capture human expertise and format it so it can be used by nonexperts.
They are rule-based systems that use if…then rules like decision trees
with the decision points for criteria provided by experts instead of
statistical analyses.
Decision trees typically have fewer than a dozen rules, whereas expert
systems can have hundreds or thousands of rules.
They gather data from experts rather than using data-mining
techniques.
They are created by interviewing experts in a given business domain
and codifying the rules stated by those experts in if…then rules.
© Pearson Prentice Hall 2009
9-33
Q6 – What are typical knowledge-management applications?
Problems
The experts are difficult to interview because they frequently have
trouble defining the rules they use.
They are difficult and expensive to develop and they divert
experts from their work and require highly skilled, expensive
developers.
They are difficult to maintain because the rules are constantly
changing, changes produce unexpected consequences, and the
rules are interdependent .
They have been unable to live up to the high expectations set by
their name.
© Pearson Prentice Hall 2009
9-34
Q6 – What are typical knowledge-management applications?
This is an example of the output from a medical expert system that is part
of a decision support system. Based on the system’s rules, an alert is
issued if the system detects a problem with a patient’s prescriptions.
Fig 9-21 Alert from Pharmacy Clinical Decision Support System
© Pearson Prentice Hall 2009
9-35
Q1 – Why do organizations need business intelligence?
Q2 – What business intelligence systems are available?
Q3 – What are typical reporting applications?
Q4 – What are typical data-mining applications?
Q5 – What is the purpose of data warehouses and data marts?
Q6 – What are typical knowledge-management applications?
Q7 – How are business intelligence applications
delivered?
© Pearson Prentice Hall 2009
9-36
Q7 – How are business intelligence applications delivered?
A data source is processed by a BI tool to produce application results.
A BI application server delivers those results in a variety of formats to
devices for consumption by BI users.
A BI server provides two functions: management and delivery.
Fig 9-22 Components of Generic Business Intelligence System
© Pearson Prentice Hall 2009
9-37
Q7 – How are business intelligence applications delivered?
The management function of a BI server maintains metadata about the
authorized allocation and timing of BI results to users. It tracks what results
are available, who is authorized to view them, and when the results are
provided to users. Here are options for managing BI results:
Users can pull their results from a Web site using a portal server with a
customizable user interface.
A server can automatically push information to users through alerts and
exception alerts, which are messages announcing events as they
occur.
A report server, a special server dedicated to reports, can supply users
with information on-demand.
© Pearson Prentice Hall 2009
9-38
Q7 – How are business intelligence applications delivered?
This figure shows a portal server that provides common data to
users. It can be used to help companies manage their knowledge.
Fig 9-23 Sample Portal, Provided by iGoogle
© Pearson Prentice Hall 2009
9-39
Q7 – How are business intelligence applications delivered?
Here are the characteristics of the delivery function of a BI
server:
It tracks authorized users.
It tracks the schedule for providing results to users.
It uses exception alerts that notify users of an exceptional
event.
The procedures used depend on the nature of the BI system.
Procedures tend to be more flexible than those in an operational
system because users of a BI system tend to be engaged in work
that is neither structured nor routine.
Procedures are determined by the unique requirements of
users.
© Pearson Prentice Hall 2009
9-40