Vous êtes sur la page 1sur 34

Management Information Systems,

Sixth Edition

Chapter 11:
Business Intelligence
and Knowledge Management
Objectives

• Explain the concepts of data mining and online


analytical processing
• Explain the notion of business intelligence and
its benefits to organizations
• Identify needs for knowledge storage and
management in organizations
• Explain the challenges in knowledge
management and its benefits to organizations

Management Information Systems, Sixth Edition 2


Objectives (continued)

• Identify possible ethical and societal issues


arising from the increasing globalization of
information technology

Management Information Systems, Sixth Edition 3


Data Mining and Online Analysis

• Data warehouse: a large database containing


historical transactions and other data
• Data warehouses are useless without software
tools to process the data into meaningful
information
• Business intelligence (BI): information gleaned
with information analysis tools
– Also called business analytics

Management Information Systems, Sixth Edition 4


Data Mining
• Data mining: the process of selecting, exploring,
and modeling large amounts of data
– Used to discover relationships or useful patterns
that can support decision making
• Data-mining tools may use complex statistical
analysis applications
• Data-mining queries are more complex than
traditional queries
• Combination of data-warehousing techniques
and data-mining tools facilitates the prediction of
future outcomes
Management Information Systems, Sixth Edition 5
Data Mining (continued)
• Data mining has four main objectives:
– Sequence or path analysis: finding patterns
where one event leads to another
– Classification: finding whether certain facts fall
into predefined groups
– Clustering: finding groups of related facts not
previously known
– Forecasting: discovering patterns that can lead
to reasonable predictions

Management Information Systems, Sixth Edition 6


Data Mining (continued)
• Data mining techniques are applied to various
fields, including marketing, fraud detection,
and targeted marketing to individuals
• Predicting customer behavior:
– Banking: help find profitable customers, detect
patterns of fraud, and predict bankruptcies
– Mobile phone services vendors: help determine
factors that affect customer loyalty
• Customer loyalty programs ensure a steady
flow of customer data into data warehouses

Management Information Systems, Sixth Edition 7


Management Information Systems, Sixth Edition 8
Data Mining (continued)

• Many industries utilize loyalty programs


– Examples include frequent-flier programs and
consumer clubs
– These programs amass huge amounts of data
about customers
• UPS has a Customer Intelligence Group
– Analyzes customer behavior
– Predicts customer defections so that a
salesperson can intervene to resolve problems

Management Information Systems, Sixth Edition 9


Data Mining (continued)
• Identifying profitable customer groups
– Financial institutions dismiss high-risk customers
– Companies attempt to define narrow groups of
potentially profitable customers
• Utilizing loyalty programs
– Amass huge amounts of data about customers
– Help companies perform yield management and
price-discrimination
– Example: Harrah’s charges higher per-night rates
to low-volume gamblers

Management Information Systems, Sixth Edition 10


Data Mining (continued)
• Inferring demographics
– Predict what customers are likely to purchase in
the future
– Amazon.com
• Determines a customer’s age range based on his
or her purchase history
• Attempts to determine customer’s gender
• Advertises for appropriate age groups based on
the inferred customer demographics
• Anticipates holidays

Management Information Systems, Sixth Edition 11


Data Mining – Machine Learning
• Machine Learning
– is an automated process that extracts patterns from
data.
– Two basic types
• Supervised ML - automatically learn a model of the
relationship between a set of input (or descriptive features)
and an output (or target features) based on a set of
historical examples, or instances.
• Unsupervised ML – is when you have an input but you don’t
have a corresponding output. The result is neither right or
wrong but the idea is to discover patterns.

Management Information Systems, Sixth Edition 12


Data Mining – Machine Learning
• Supervised ML (an example)
– Below is an a historical examples or instances of loans data of
customers from a bank. The descriptive features of this dataset
are Occupation, Age and Loan-Salary Ratio and its target feature
is the Outcome field.
ID Occupation Age Loan-Salary Ration Outcome
1 Industrial 34 2.96 Repay
2 Professional 41 4.64 Default
3 Professional 36 3.22 Default
4 Professional 41 3.11 Default
5 Industrial 48 3.80 Default
6 Industrial 61 2.52 Repay
7 Professional 37 1.50 Repay
8 Professional 40 1.93 Repay
9 Industrial 33 5.25 Default
10 Industrial 32 4.15 Default
Management Information Systems, Sixth Edition 13
Data Mining – Machine Learning
• Supervised ML (an example)
– Based on the historical instances as shown on the previous table.
The descriptive features (i.e. Age, Occupation and Loan-Salary
Ratio) map into the target feature (i.e. Outcome), namely, whether
a bank customer defaulted (did not pay the loan) or was able to
repay the loan, the machine learn the following model:

If (Loan-Salary-Ratio > 3) then


Outcome = Default
Else
Outcome = Repay

Management Information Systems, Sixth Edition 14


Data Mining – Machine Learning
• Flowchart for the model

T
If (Loan-Salary-Ratio > 3) Print Repay

Print Default

Management Information Systems, Sixth Edition 15


Data Mining – Machine Learning
• Example C++ code:
#include <iostream>
#include <string>
#include <iomanip>
using namespace std;
using std::setw;
int main () {
int age[] = {34,41,36,41,48,61,37,40,33,32};
double loan_salary_ratio[] = {2.96, 4.64, 3.22, 3.11, 3.80, 2.52,
1.50, 1.93, 5.25, 4.15};
string occupation[] = {"Industrial", "Professional", "Professional",
"Professional", "Industrial","Industrial","Professional" ,"Professional"
,"Industrial","Industrial"};

cout << "ID" << setw(10) << "Occupation" << setw(14) << "Age" << setw(12)
<< "Loan Salary Ratio" << setw(20) << "Outcome" << endl;

for ( int i = 0; i < 10; i++ ) {


cout << i << setw(7) << occupation[i] << setw(20) << age[i] <<
setw(12) << loan_salary_ratio[i];
if (loan_salary_ratio[i]>3)
cout << setw(25) << "Repay" << endl;
else
cout << setw(25) << "Default" << endl;
}
return 0;
}

Management Information Systems, Sixth Edition 16


Data Mining – Machine Learning
• Data Mining Activity Part I – Below also is another
historical examples or instances of banks loans from
different customers.
ID Occupation Age Loan-Salary-Ratio Outcome
1 Industrial 44 3.69 Repay
2 Industrial 41 1.20 Repay
3 Industrial 37 3.75 Default
4 Industrial 44 2.33 Repay
5 Professional 39 4.21 Default
6 Industrial 29 2.94 Default
7 Professional 38 2.64 Repay
8 Professional 17 2.77 Repay
9 Professional 30 1.33 Repay
10 Industrial 30 3.78 Default

Management Information Systems, Sixth Edition 17


Data Mining – Machine Learning
• Supervised ML (an example)
– Using the descriptive features of Loan-Salary Ratio, Age and Occupation
mapping to Target feature (i.e. Outcome field) which has either the
values of Repay or Default, create a Model and a Flowchart later on.
– So, imagine yourself as the software (or algorithm for that matter) for
machine learning. What possible pattern from the previous table could
you detect that you could turn into a Model?
– Your goal is to produce a Model (see Slide 14) and a Flowchart of the
model (see Slide 15).
– This is worth 20 points for the Model and 20 points also for the Flowchart.
– Open the file that I send you and write your name (after Student Name).
Make your Model and Flowchart and once your done send it to
nationalcis@gmail.com with e-mail title Data Mining Activity Part I.
– Tip : If you look at the Model on Slide 14 it has one (actually including the else
statement) pattern, namely, if a loan has loan-salary ration greater than 3 then the
Loan is not paid (or on Default) otherwise the customer was able to Repay. Now,
on this exercise there are three possible patterns (and again four if you include the
else statement). So, your first task is to detect patterns from the table on Slide 17
and turn it into a Model. And after the Model is done make a flowchart out of it.

Management Information Systems, Sixth Edition 18


Data Mining – Machine Learning
• Data Mining Activity Part II – Convert your flowchart (which is
based on your Model) into a C++ program. See Slides 15 and 16 for
example.
• Open the MSWord document that I send you and write your name
after Student Name label.
• Open your Visual Studio and create a Project in C++ with a name
Data Mining Your Last Name (e.g. Data Mining Smith). Erase all the
code. And then open the sample code that I send you in Notepad and
then copy paste its content into your Visual Studio.
• Convert your Flowchart into C++ code (See Slides 15 and 16 as your
example for that) and make sure that it runs and it has no bug.
• Once you are completed, paste your code from Visual Studio on the
Microsoft Word document that I send you and then save it. Then send
it to nationalcis@gmail.com with title Data Mining Activity Part II.

Management Information Systems, Sixth Edition 19


Online Analytical Processing
• Online analytical processing (OLAP): a type
of application used to exploit data warehouses
– Provides extremely fast response times
– Allows a user to view multiple combinations of
two dimensions by rotating virtual “cubes” of
information
• Drilling down: the process of starting with broad
information and then retrieving more specific
information as numbers or percentages
• Can use relational or dimensional databases
designed for OLAP applications
Management Information Systems, Sixth Edition 20
Management Information Systems, Sixth Edition 21
Online Analytical Processing
(continued)
• OLAP application composes tables “on the fly”
based on the desired relationships
• Dimensional database: data is organized into
tables showing information summaries
– Also called multidimensional databases
• OLAP applications are powerful tools for
executives

Management Information Systems, Sixth Edition 22


Management Information Systems, Sixth Edition 23
Online Analytical Processing
(continued)
• Ruby Tuesday restaurant chain case
– One location was performing below average
– OLAP analysis showed that customers were
waiting longer than normal
– Appropriate changes were made
• OLAP applications are usually installed on a
special server
• OLAP applications are usually significantly faster
than relational applications

Management Information Systems, Sixth Edition 24


Management Information Systems, Sixth Edition 25
Online Analytical Processing
(continued)
• OLAP is increasingly used by corporations to
gain efficiencies
– Office Depot used OLAP on a data warehouse to
determine cross-selling strategies
– Ben & Jerry’s tracks ice cream flavor popularity
• BI software is becoming easier to use
– Intelligent interfaces accept queries in free form
• BI software is integrated into Microsoft’s SQL
Server database software

Management Information Systems, Sixth Edition 26


More Customer Intelligence
• A major effort of business is collecting business
intelligence about customers
• Data-mining and OLAP software are often
integrated into CRM systems
• Web has become popular for transactions,
making data collection easy
• Targeted marketing is more effective than mass
marketing
• Clickstream software: tracks and stores data
about every visit to a Web site
Management Information Systems, Sixth Edition 27
Management Information Systems, Sixth Edition 28
More Customer Intelligence (continued)
• Data from customer activity on a Web site may
not provide a full picture
• Third-party companies such as DoubleClick and
Engage Software may be hired to study
consumer activity
– These companies compile billions of consumer
clickstreams to create behavioral models
• Can determine consumers’ interests by
capturing where, what, when, and how often
Web pages are visited, ads are clicked, and
transactions are completed
Management Information Systems, Sixth Edition 29
More Customer Intelligence (continued)

• Drugstore.com: a Web-based drugstore


– Wanted to reach more customers
– Hired Avenue A | Razorfish Inc. to do customer
profiling
• Avenue A compiles anonymous information
about customers continuously, and also
collected and analyzed data from Drugstore.com
– Discovered basic themes in shopper behavior
that will help Drugstore.com determine where and
how to advertise to gain new customers
Management Information Systems, Sixth Edition 30
Dashboards

• Dashboard: an interface between BI tools and


the user
– Resembles a car dashboard
– Contains visual images to quickly represent
specific business metrics of interest to
management
– Helps management monitor revenue and sales,
monitor inventory levels, and pinpoint trends and
changes over time

Management Information Systems, Sixth Edition 31


Summary
• Business intelligence (BI) is any information
about organization, its customers, or its
suppliers that can help firms make decisions
• Data mining is the process of selecting,
exploring, and modeling large amounts of data
to discover previously unknown relationships
• Data mining is useful for predicting customer
behavior and detecting fraud
• Online analytical processing (OLAP) puts data
into two-dimensional tables
Management Information Systems, Sixth Edition 32
Summary (continued)
• OLAP either uses dimensional databases or
calculates desired tables on the fly
• Drilling down means moving from a broad view
to a specific view of information
• Dashboards interface with BI software tools to
provide quick information such as business
metrics
• Knowledge management involves gathering,
organizing, sharing, analyzing, and
disseminating knowledge

Management Information Systems, Sixth Edition 33


Summary (continued)

• The main challenge of knowledge management


is identifying and classifying useful information
from unstructured sources
• Most unstructured knowledge is textual
• Employee knowledge networks are software
tools to help employees find other employees
with specific expertise
• Autocategorization is the automatic classification
of information

Management Information Systems, Sixth Edition 34

Vous aimerez peut-être aussi