Transcript Introduction to Data Mining

Introduction to Data Mining
Why Mine Data? Commercial Viewpoint
• Lots of data is being collected
and warehoused
– Web data, e-commerce
– purchases at department/
grocery stores
– Bank/Credit Card
transactions
• Twice as much information was created in 2002 as in 1999 (~30% growth rate)
• Other growth rate estimates even higher
Largest databases in 2007
• Largest database in the world: World Data Centre for Climate
(WDCC) operated by the Max Planck Institute and German
Climate Computing Centre
– 220 terabytes of data on climate research and climatic trends,
– 110 terabytes worth of climate simulation data.
– 6 petabytes worth of additional information stored on tapes.
• AT&T
– 323 terabytes of information
– 1.9 trillion phone call records
• Google
– 91 million searches per day,
• After a year worth of searches, this figure amounts to more than 33
trillion database entries.
Why Mine Data? Scientific Viewpoint
• Data is collected and stored at
enormous speeds (GB/hour). E.g.
– remote sensors on a satellite
– telescopes scanning the skies
– scientific simulations
generating terabytes of data
• Very little data will ever be looked at
by a human
• Knowledge Discovery is NEEDED
to make sense and use of data.
Data Mining
• Data mining is the process of automatically discovering useful
information in large data repositories.
• Human analysts may take weeks to discover useful information.
• Much of the data is never analyzed at all.
4,000,000
3,500,000
The Data Gap
3,000,000
2,500,000
2,000,000
1,500,000
Total new disk (TB) since 1995
1,000,000
Number of
analysts
500,000
0
1995
1996
1997
1998
1999
From: R. Grossman, C. Kamath, V. Kumar, “Data Mining for Scientific and Engineering Applications”
What is (not) Data Mining?
What is not Data
Mining?


What is Data Mining?
– Look up phone
number in phone
directory
– Certain names are more
prevalent in certain locations
(O’Brien, O’Rurke, O’Reilly…
in Boston area)
– Query a Web
search engine for
information about
“Amazon”
–Discover groups of similar
documents on the Web
Origins of Data Mining
• Draws ideas from: machine learning/AI, statistics, and database
systems
Statistics
Machine Learning
Data Mining
Database
systems
Data Mining Tasks
Data mining tasks are generally divided into two major categories:
• Predictive tasks [Use some attributes to predict unknown or future
values of other attributes.]
• Classification
• Regression
• Deviation Detection
• Descriptive tasks [Find human-interpretable patterns that describe the
data.]
• Association Discovery
• Clustering
Predictive Data Mining or
Supervised learning
• Given a collection of records (training set)
– Each record contains a set of attributes, one of the attributes is the class.
• Find ("learn") a model for the class attribute as a function of the
values of the other attributes.
• Goal: previously unseen records should be assigned a class as
accurately as possible.
Learning
We can think of at least three different problems being involved in
learning:
• memory,
• averaging, and
• generalization.
Example problem
(Adapted from Leslie Kaelbling's example in the MIT courseware)
• Imagine that I'm trying predict whether my neighbor is going to
drive into work, so I can ask for a ride.
• Whether she drives into work seems to depend on the following
attributes of the day:
–
–
–
–
temperature,
expected precipitation,
day of the week,
what she's wearing.
Memory
• Okay. Let's say we observe our neighbor on three days:
Temp
Precip
Day
Shop
Clothes
25
None
Sat
No
Casual
Walk
-5
Snow
Mon
Yes
Casual
Drive
15
Snow
Mon
Yes
Casual
Walk
Memory
• Now, we find ourselves on a snowy “–5” – degree Monday, when
the neighbor is wearing casual clothes and going shopping.
• Do you think she's going to drive?
Temp
Precip
Day
Clothes
25
None
Sat
Casual
Walk
-5
Snow
Mon
Casual
Drive
15
Snow
Mon
Casual
Walk
-5
Snow
Mon
Casual
Memory
• The standard answer in this case is "yes".
– This day is just like one of the ones we've seen before, and so it
seems like a good bet to predict "yes."
• This is about the most rudimentary form of learning, which is just
to memorize the things you've seen before.
Temp
Precip
Day
Clothes
25
None
Sat
Casual
Walk
-5
Snow
Mon
Casual
Drive
15
Snow
Mon
Casual
Walk
-5
Snow
Mon
Casual
Drive
Noisy Data
• Things aren’t always as easy as they were in the previous case. What if
you get this set of noisy data?
Temp
Precip
Day
Clothes
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Drive
25
None
Sat
Casual
Drive
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
?
• Now, we are asked to predict what's going to happen.
• We have certainly seen this case before.
• But the problem is that it has had different answers. Our neighbor is
not entirely reliable.
Averaging
• One strategy would be to predict the majority outcome.
– The neighbor walked more times than she drove in this situation, so
we might predict "walk".
Temp
Precip
Day
Clothes
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Drive
25
None
Sat
Casual
Drive
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
25
None
Sat
Casual
Walk
Generalization
• Dealing with previously unseen cases
• Will she walk or drive?
Temp
Precip
Day
Clothes
22
None
Fri
Casual
Walk
3
None
Sun
Casual
Walk
10
Rain
Wed
Casual
Walk
30
None
Mon
Casual
Drive
20
None
Sat
Formal
Drive
25
None
Sat
Casual
Drive
-5
Snow
Mon
Casual
Drive
27
None
Tue
Casual
Drive
24
Rain
Mon
Casual
?
• We might plausibly
make any of the
following arguments:
– She's going to
walk because it's
raining today and
the only other time
it rained, she
walked.
– She's going to
drive because she
has always driven
on Mondays…
Classification Another Example
Tid Refund Marital
Status
Taxable
Income Cheat
Refund Marital
Status
Taxable
Income Cheat
1
Yes
Single
125K
No
No
Single
75K
?
2
No
Married
100K
No
Yes
Married
50K
?
3
No
Single
70K
No
No
Married
150K
?
4
Yes
Married
120K
No
Yes
Divorced 90K
?
5
No
Divorced 95K
Yes
No
Single
40K
?
6
No
Married
No
No
Married
80K
?
60K
10
7
Yes
Divorced 220K
No
8
No
Single
85K
Yes
9
No
Married
75K
No
10
10
No
Single
90K
Yes
Training
Set
Learn
Classifier
Test
Set
Model
Example of a Decision Tree
Tid Refund Marital
Status
Taxable
Income Cheat
1
Yes
Single
125K
No
2
No
Married
100K
No
3
No
Single
70K
No
4
Yes
Married
120K
No
5
No
Divorced 95K
Yes
6
No
Married
No
7
Yes
Divorced 220K
No
8
No
Single
85K
Yes
9
No
Married
75K
No
10
No
Single
90K
Yes
60K
Splitting Attributes
Refund
Yes
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
NO
> 80K
YES
10
Training Data
Married
Model: Decision Tree
Apply Model to Test Data
Test Data
Start from the root of tree.
Refund
Yes
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
Apply Model to Test Data
Test Data
Refund
Yes
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
Apply Model to Test Data
Test Data
Refund
Yes
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
Apply Model to Test Data
Test Data
Refund
Yes
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
Apply Model to Test Data
Test Data
Refund
Yes
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
Apply Model to Test Data
Test Data
Refund
Yes
Refund Marital
Status
Taxable
Income Cheat
No
80K
Married
?
10
No
NO
MarSt
Single, Divorced
TaxInc
< 80K
NO
Married
NO
> 80K
YES
Assign Cheat to “No”
Classification: Direct Marketing
– Goal: Reduce cost of mailing by targeting a set of consumers likely
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, don’t buy} decision forms the class attribute.
• Collect various demographic, lifestyle, and other related information
about all such customers. E.g.
• Type of business,
• where they stay,
• how much they earn, etc.
• Use this information as input attributes to learn a classifier model.
Classification: Fraud Detection
• Goal: Predict fraudulent cases in credit
card transactions.
• Approach:
• Use credit card transactions and the
information associated with them as
attributes, e.g.
– when does a customer buy,
– what does he buy,
– where does he buy, etc.
• Label some past transactions as fraud or fair
transactions. This forms the class attribute.
• Learn a model for the class of the
transactions.
• Use this model to detect fraud by observing
credit card transactions on an account.
Classification: Attrition/Churn
• Situation: Attrition rate for mobile
phone customers is around 25-30% a
year!
• Goal: To predict whether a customer is
likely to be lost to a competitor.
• Approach:
Success story (Reported in 2003):
• Verizon Wireless performed
this kind of data mining
• Use detailed record of transactions with
reducing attrition rate from
each of the past and present customers, to
over 2% per month to under
find attributes. E.g.
1.5% per month.
– how often the customer calls,
• Huge impact, with >30 M
– where he calls,
subscribers (0.5% is 150,000
– what time-of-the day he calls most,
customers).
– his financial status,
– marital status, etc.
• Label the customers as loyal or disloyal.
Find a model for loyalty.
Assessing Credit Risk
• Situation: Person applies for a loan
• Task: Should a bank approve the loan?
– People who have the best credit don’t need the loans
– People with worst credit are not likely to repay.
– Bank’s best customers are in the middle
• Banks develop credit models using a variety of data mining
methods.
• Mortgage and credit card proliferation are the results of being
able to "successfully" predict if a person is likely to default on a
loan.
• Widely deployed in many countries.
Frequent-Itemset Mining
(Association Discovery)
The Market-Basket Model
• A large set of items, e.g., things sold in a supermarket.
• A large set of baskets, each of which is a small set of the items,
e.g., the things one customer buys on one day.
Fundamental problem
• What sets of items are often bought together?
Application
• If a large number of baskets contain both hot dogs and mustard,
we can use this information in several ways. How?
Hot Dogs and Mustard
1. Apparently, many people walk from where the hot dogs are to
where the mustard is.
•
•
We can put them close together, and put between them other foods
that might also be bought with hot dogs and mustard, e.g., ketchup
or potato chips.
Doing so can generate additional "impulse" sales.
2. The store can run a sale on hot dogs and at the same time raise
the price of mustard.
•
•
•
People will come to the store for the cheap hot dogs, and many will
need mustard too.
It is not worth the trouble to go to another store for cheaper
mustard, so they buy that too.
The store makes back on mustard what it loses on hot dogs, and
also gets more customers into the store.
Beer and Diapers
• What’s the explanation here?
On-Line Purchases
• Amazon.com offers several million different items for sale, and
has several tens of millions of customers.
• Basket = Customer, Item = Book, DVD, etc.
– Motivation: Find out what items are bought together.
• Basket = Book, DVD, etc. Item = Customer
– Motivation: Find out similar customers.
Words and Documents
• Baskets = sentences; items = words in those sentences.
– Lets us find words that appear together unusually frequently, i.e.,
linked concepts.
• Baskets = sentences, items = documents containing those
sentences.
– Items that appear together too often could represent plagiarism.
Genes
• Baskets = people; items = genes or blood-chemistry factors.
– Has been used to detect combinations of genes that result in
diabetes
Clustering
• 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.
– Data points in separate clusters are less similar to one another.
• Similarity Measures:
– Euclidean Distance if attributes are continuous.
– Other Problem-specific Measures.
E.g. Euclidean Distance Based Clustering in 3-D space.
Intracluster distances
are minimized
Intercluster distances
are maximized
Clustering: Application 1
• Market Segmentation:
– Goal: subdivide a market into distinct subsets of customers
where any subset may conceivably be selected as a market
target to be reached with a distinct marketing mix.
– Approach:
• Collect different attributes of customers based on their
geographical and lifestyle related information.
• Find clusters of similar customers.
Clustering: Application 2
• Document Clustering:
– Goal: To find groups of documents that are similar to each other based
on the important words appearing in them.
– Approach:
• Identify frequently occurring words 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 to clustered documents.
There are two natural
clusters in the data set.
The first cluster
consists of the first four
articles, which
correspond to news
about the economy.
The second cluster
contains the last four
articles, which
correspond to news
about health care.
Each article is represented as a set of wordfrequency pairs (w, c).