2a. basic data mining techniques

8/2/2019 2a. Basic Data Mining Techniques

1/39

Basic Data Mining Techniques


2/39

Prediction Methods

Use some variables to predict unknown or future values of

other variables.

Description Methods

Find human-interpretable patterns that describe the data.

From [Fayyad, et.al.] Advances in Knowledge Discovery and Data Mining, 1996


3/39

Classification [Predictive]

Deviation Detection [Predictive]

Association Rule Discovery [Descriptive]


4/39

Given a collection of records (training set ),

the class. Find a model for class attribute as a function of the

va ues o ot er attr utes.

Goal: previously unseen records should be assigned aclass as accuratel as ossible. A test set is used to determine the accuracy of the model.

Usually, the given data set is divided into training and test sets,with training set used to build the model and test set used to

validate it.


5/39

Tid Refund MaritalStatus

TaxableIncome Cheat

1 Yes Single 125K No

Refund Marital

Status

Taxable

Income Cheat

No Single 75K ?

2 No Married 100K No

3 No Single 70K No

4 Yes Married 120K No

Yes Married 50K ?

No Married 150K ?

Yes Divorced 90K ?

o vorce es

6 No Married 60K No

7 Yes Divorced 220K No

o ng e 40

No Married 80K ?10

TestSet

9 No Married 75K No

10 No Single 90K Yes10

TrainingSet

ModelLearn

Classifier


6/39

Direct Marketing

to buy a new cell-phone product. Approach:

Use the data for a similar product introduced before.

We know which customers decided to buy and which decided

otherwise. This {buy, dont buy} decision forms the class attribute. o ect various emograp ic, i esty e, an company-interaction re ate

information about all such customers.

Type of business, where they stay, how much they earn, etc.

se s n orma on as npu a r u es o earn a c ass er mo e .

From [Berry & Linoff] Data Mining Techniques, 1997


7/39

Fraud Detection

.

Approach: Use credit card transactions and the information on its account-holder

as attributes.

When does a customer buy, what does he buy, how often he payson time, etc

Label past transactions as fraud or fair transactions. This forms the classa r u e.

Learn a model for the class of the transactions.

Use this model to detect fraud by observing credit card transactions onan account.


8/39

Customer Attrition/Churn:

competitor. Approach:

Use detailed record of transactions with each of the past and

present customers, to find attributes.

How often the customer calls, where he calls, what time-of-the day he calls most, his financial status, marital status, etc.

Label the customers as loyal or disloyal.

Find a model for loyalty.

From [Berry & Linoff] Data Mining Techniques, 1997


9/39

Sky Survey Cataloging

Goal: To redict class star or alax of sk ob ects, es eciall

visually faint ones, based on the telescopic survey images (fromPalomar Observatory).

3000 images with 23,040 x 23,040 pixels per image.

Approach:

Segment the image.- .

Model the class based on these features.

Success Story: Could find 16 new high red-shift quasars, some of the

farthest ob ects that are difficult to find!

From [Fayyad, et.al.] Advances in Knowledge Discovery and Data Mining, 1996


10/39

Given a set of data points, each having a set of attributes,

and a similarity measure among them, find clusters such

that Data points in one cluster are more similar to one another.

another.

Similarity Measures: Euclidean Distance if attributes are continuous.

Other Problem-specific Measures.


11/39

Euclidean Distance Based Clustering in 3-D space.

Intracluster distancesare minimized

Intracluster distancesare minimized

Intercluster distancesare maximized

Intercluster distancesare maximized


12/39

Market Segmentation:

Goal: subdivide a market into distinct subsets of customers whereany subset may conceivably be selected as a market target to bereached with a distinct marketing mix.

Approach:

Collect different attributes of customers based on their geographical andlifestyle related information.

Find clusters of similar customers. easure t e c uster ng qua ty y o serv ng uy ng patterns o customers

in same cluster vs. those from different clusters.


13/39

Document Clustering:

other based on the important terms appearing in them. Approach: To identify frequently occurring terms in each

document. Form a similarity measure based on the

frequencies of different terms. Use it to cluster.

Gain: Information Retrieval can utilize the clusters to relate a

new document or search term to clustered documents.


14/39

Clustering Points: 3204 Articles of Los Angeles Times.

Similarit Measure: How man words are common in these

documents (after some word filtering).

Ar t i cl es Pl aced

Financial 555 364

National 273 36

Sports 738 573

Entertainment 354 278


15/39

Observe Stock Movements every day. Clustering points: Stock-{UP/DOWN}

described by them frequently happen together on the same day.We used association rules to quantify a similarity measure.

1Applied-Matl-DOWN,Bay-Network-Down,3-COM-DOWN,

Cabletron-Sys-DOWN,CISCO-DOWN,HP-DOWN,

DSC-Comm-DOW N,INTEL-DOWN,LSI-Logic-DOWN,

Micron-Tech-DOWN,Texas-Inst-Down,Tellabs-Inc-Down,Natl-Semiconduct-DOWN,Oracl-DOWN,SGI-DOWN,

Technology1-DOWN

Sun-DOW N

2Apple-Comp-DOWN,Autodesk-DOWN,DEC-DOWN,

ADV-Micro-Device-DOWN,Andrew-Corp-DOWN,

Computer-Assoc-DOWN,Circuit-City-DOWN,

Compaq-DOWN, EMC-Corp-DOWN, Gen-Ins t-DOWN,

Motorola-DOW N,Microsoft-DOWN,Scientific-Atl-DOWN

Technology2-DOWN

3Fannie-Mae-DOWN,Fed-Home-Loan-DOWN,

MBNA-Corp-DOWN,Morgan-Stanley-DOWN Financial-DOWN

4Baker-Hughes-UP,Dresser-Inds-UP,Halliburton-HLD-UP,

Louisiana-Land-UP,Phillips-Petro-UP,Unocal-UP,

Schlumberger-UPOil-UP


16/39

Given a set of records each of which contain some number of items

from a given collection;

Produce dependency rules which will predict occurrence of an item

based on occurrences of other items.

TID Items

1 Bread, Coke, Milk

2 Beer, Bread

3 Beer, Coke, Diaper, Milk

4 Beer, Bread, Diaper, Milk

{ Milk } - -> { Coke}

{ Diaper, Mi lk} - -> { Beer}

{ Milk } - -> { Coke}

{ Diaper, Mi lk} - -> { Beer}

5 Coke, Diaper, Milk


17/39

Marketing and Sales Promotion:

{Bagels, } --> {Potato Chips} Potato Chips as consequent => Can be used to determine what

.

Bagels in the antecedent => Can be used to see which products

would be affected if the store discontinues selling bagels.used to see what products should be sold with Bagels to promotesale of Potato chips!


18/39

Supermarket shelf management.

many customers. Approach: Process the point-of-sale data collected with

barcode scanners to find dependencies among items.

A classic rule --

If a customer bu s dia er and milk then he is ver likel to bu

beer.

So, dont be surprised if you find six-packs stacked next to diapers!


19/39

Association Rule Discovery: Application 3

Inventory Management:

Goal: A consumer a liance re air com an wants to antici ate the

nature of repairs on its consumer products and keep the service

vehicles equipped with right parts to reduce on number of visits to

consumer households.

Approach: Process the data on tools and parts required in previous

repairs at different consumer locations and discover the co-

occurrence atterns.


20/39

Sequential Pattern Discovery: Definition

Given is a set ofobjects, with each object associated with its own timeline of events, findrules that predict strong sequential dependencies among different events.

(A B) (C) (D E)

Rules are formed by first disovering patterns. Event occurrences in the patterns aregoverned by timing constraints.

(A B) (C) (D E)


21/39

Sequential Pattern Discovery: Examples

In telecommunications alarm logs,

(Inverter_Problem Excessive_Line_Current)

(Rectifier_Alarm) --> (Fire_Alarm)

In point-of-sale transaction sequences,

Computer Bookstore:

(Intro_To_Visual_C) (C++_Primer) -->

(Perl_for_dummies,Tcl_Tk)

Athletic Apparel Store:(Shoes) (Racket, Racketball) --> (Sports_Jacket)


22/39

Predict a value of a given continuous valued variable based on the

values of other variables, assuming a linear or nonlinear model of

dependency.

Greatly studied in statistics, neural network fields.

Predicting sales amounts of new product based on advetising

expenditure. Predicting wind velocities as a function of temperature, humidity,

air pressure, etc.

Time series prediction of stock market indices.


23/39

Deviation/Anomaly Detection

Detect significant deviations from normal behavior

Applications:

Credit Card Fraud Detection

Network Intrusion

Detection

Typical network traffic at University level may reach over 100 million connections per day


24/39

Classical View

Exemplar View

To help categorize the data mining technique

Roiger, 2003


25/39

All concepts have definite defining properties.

Example:

annual income >= 30,000

& years at current position >= 5 owns ome = true

Then

Good Credit Risk = true


26/39

Concepts are represented by properties that are

probable of concept number

Example

A good credit risk might look like:

The mean annual income for individuals who consistently make loan

payments on time is $30,000Most individuals who are good credit risks have been working forthe same company for at least 5 years

The majority of good credit risks own their own home

omeowner wit an annua incomeo , emp oye at t esame position for 4 years might be classified as a good credit riskwith a probability of 0.85


27/39

People store and recall likely concept exemplars that are

Exemplar#1

Annual income = 32,000

Number of years at current position = 6

Exemplar#2

Annual income = 52,000

Number of years at current position = 16


28/39

Supervised Learning

.

Use the model to determine the outcome new instances ofunknown origin

Unsupervised Learning

ata m n ng met o t at u s mo e s rom ata w t out

predefined classes.


29/39

Supervised Learning:


30/39

Decision Tree

A tree structure where non-terminal nodes

terminal nodes reflect decision outcomes.


31/39

Table 1.1 H othetical Trainin Data for Disease Dia nosis

Patient Sore SwollenID# Throat Fever Glands Congestion Headache Diagnosis

1 Yes Yes Yes Yes Yes Strep throat2 No No No Yes Yes Allergy3 Yes Yes No Yes No Cold

4 Yes No Yes No No Strep throat5 No Yes No Yes No Cold6 No No No Yes No Allergy7 No No Yes No No Strep throat8 Yes No No Yes Yes Allergy9 No Yes No Yes Yes Cold

Input Attributes Output Attributes


32/39

Swollen

Glands

No Yes

Fever

Diagnosis = Strep Throat

Yes

Diagnosis = Allergy Diagnosis = Cold

No


33/39

Table 1.2 Data Instances with an Unknown Classification

Patient Sore SwollenID# Throat Fever Glands Congestion Headache Diagnosis

11 No No Yes Yes Yes ?12 Yes Yes No No Yes ?13 No No No No Yes ?


34/39

Translation from decision tree

Form:

IF antecedant conditions

THEN consequent conditions

Example

IF Swollen Glands = YesTHEN Diagnosis = Strep Throat

IF Swollen Glands = No & Fever = Yes

THEN Diagnosis = Cold

IF Swollen Glands = No & Fever = NoTHEN Diagnosis = Allergy


35/39

The Acme Investors Dataset & Supervised

1. Can I develop a general profile of an online investor?

2. Can I determine if a new customer is likely to open a margin account?

3. an u a mo e pre c e average num er o ra es per mon or

a new investor?4. What characteristics differentiate female and male investors?

Output attribute:No 1. transaction method

No 2. margin accountNo 3. trades/month

No 4. sex


36/39

The Acme Investors Dataset & Unsupervised

1. What attribute similarities group customers of Acme

2. What differences in attribute values segment thecustomer database?


37/39

Require us to provide an initial best estimate about the

OR Use an al orithm to determine a best number of clusters

In either case, a clusterin s stem will attem t to rouinstances into clusters of significant interests


38/39


39/39

Dunham chapter 1.2

2a. basic data mining techniques

Documents