Download Data analysis: an introduction

Introduc)on to Data Analysis and Explora)on (from the Lecture Notes of the book:
Introduction to Data Mining, by Tan,Steinbach
and Kumar)
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Analysis of data (from Wikipedia)
•  Data analysis is a process of inspec'ng, cleaning, transforming, and modeling data with the goal of highligh)ng useful informa)on, sugges)ng conclusions, and suppor)ng decision making. •  Data analysis has mul)ple facets and approaches, encompassing diverse techniques under a variety of names, in different business, science, and social science domains. •  Data explora)on: a preliminary explora)on of the data to beFer understand its characteris)cs. Now developing into richer paradigms. •  Data mining is a par)cular data analysis technique that focuses on modeling and knowledge discovery for predic)ve rather than purely descrip)ve purposes. •  Business intelligence covers data analysis that relies heavily on aggrega)on, focusing on business informa)on. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Why Analyze Data? Commercial Viewpoint •  Lots of data is being collected and warehoused –  Web data, e-­‐commerce –  purchases at department/ grocery stores –  Bank/Credit Card transac)ons •  Computers have become cheaper and more powerful •  Compe))ve Pressure is Strong –  Provide beFer, customized services for an edge (e.g. in Customer Rela)onship Management) © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Why Analyze Data? Scien)fic Viewpoint •  Data collected and stored at enormous speeds (GB/hour) –  remote sensors on a satellite –  telescopes scanning the skies –  microarrays genera)ng gene expression data –  scien)fic simula)ons genera)ng terabytes of data •  Tradi)onal techniques infeasible for raw data •  Data mining may help scien)sts –  in classifying and segmen)ng data –  in Hypothesis Forma)on © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Analyzing Large Data Sets -­‐ Mo)va)on •  There is o)en informa.on “hidden” in the data that is not readily evident •  Human analysts may take weeks to discover useful informa.on •  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:
Tan,Steinbach,
R. Grossman,
Kumar
C. Kamath, V. Kumar,
Introduction
“Data Mining
to Data
for Mining
Scientific and Engineering Applications”
8/05/2005
‹#›
What is data explora)on? A preliminary exploration of the data to
better understand its characteristics.
•  Key mo)va)ons of data explora)on include –  Helping to select the right tool for preprocessing or analysis –  Making use of humans’ abili)es to recognize paFerns •  People can recognize paFerns not captured by data analysis tools •  Related to the area of Exploratory Data Analysis (EDA) –  Created by sta)s)cian John Tukey –  Seminal book is Exploratory Data Analysis by Tukey –  A nice online introduc)on can be found in Chapter 1 of the NIST Engineering Sta)s)cs Handbook hFp://www.itl.nist.gov/div898/handbook/index.htm © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Techniques Used In Data Explora)on •  In EDA, as originally defined by Tukey –  The focus was on visualiza)on –  Clustering and anomaly detec)on were viewed as exploratory techniques –  In data mining, clustering and anomaly detec)on are major areas of interest, and not thought of as just exploratory •  Basic tradi)onal techniques of data explora)on –  Summary sta)s)cs –  Visualiza)on –  Online Analy)cal Processing (OLAP) © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Summary Sta)s)cs •  Summary sta)s)cs are numbers that summarize proper)es of the data –  Summarized proper)es include frequency, loca)on and spread •  Examples: loca)on -­‐ mean spread -­‐ standard devia)on –  Most summary sta)s)cs can be calculated in a single pass through the data © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Frequency and Mode •  The frequency of an aFribute value is the percentage of )me the value occurs in the data set –  For example, given the aFribute ‘gender’ and a representa)ve popula)on of people, the gender ‘female’ occurs about 50% of the )me. •  The mode of a an aFribute is the most frequent aFribute value •  The no)ons of frequency and mode are typically used with categorical data © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Percen)les •  For con)nuous data, the no)on of a percen)le is more useful. Given an ordinal or con)nuous aFribute x and a number p between 0 and 100, the pth percen)le is a value x p of x such that p% of the observed values of x are less than x p € instance, the 50th percen)le x 50%
•  For is the value such that 50% o€
f all values of x are less than x 50%
xp
€
© Tan,Steinbach, Kumar
Introduction to Data Mining
€
8/05/2005
‹#›
Measures of Loca)on: Mean and Median •  The mean is the most common measure of the loca)on of a set of points. •  However, the mean is very sensi)ve to outliers. •  Thus, the median or a trimmed mean is also commonly used. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Measures of Spread: Range and Variance •  Range is the difference between the max and min •  The variance or standard devia)on is the most common measure of the spread of a set of points. •  However, this is also sensi)ve to outliers, so that other measures are oden used. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Visualiza)on Visualiza)on is the conversion of data into a visual or tabular format so that the characteris)cs of the data and the rela)onships among data items or aFributes can be analyzed or reported. •  Visualiza)on of data is one of the most powerful and appealing techniques for data explora)on. –  Humans have a well developed ability to analyze large amounts of informa)on that is presented visually –  Can detect general paFerns and trends –  Can detect outliers and unusual paFerns © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Example: Sea Surface Temperature •  The following shows the Sea Surface Temperature (SST) for July 1982 –  Tens of thousands of data points are summarized in a single figure © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Representa)on •  Is the mapping of informa)on to a visual format •  Data objects, their aFributes, and the rela)onships among data objects are translated into graphical elements such as points, lines, shapes, and colors. •  Example: –  Objects are oden represented as points –  Their aFribute values can be represented as the posi)on of the points or the characteris)cs of the points, e.g., color, size, and shape –  If posi)on is used, then the rela)onships of points, i.e., whether they form groups or a point is an outlier, is easily perceived. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Selec)on •  Is the elimina)on or the de-­‐emphasis of certain objects and aFributes •  Selec)on may involve choosing a subset of aFributes –  Dimensionality reduc)on is oden used to reduce the number of dimensions to two or three –  Alterna)vely, pairs of aFributes can be considered •  Selec)on may also involve choosing a subset of objects –  A region of the screen can only show so many points –  Can sample, but want to preserve points in sparse areas © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Visualiza)on Techniques: Histograms •  Histogram –  Usually shows the distribu)on of values of a single variable –  Divide the values into bins and show a bar plot of the number of objects in each bin. –  The height of each bar indicates the number of objects –  Shape of histogram depends on the number of bins •  Example: Petal Width (10 and 20 bins, respec)vely) © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Two-­‐Dimensional Histograms •  Show the joint distribu)on of the values of two aFributes •  Example: petal width and petal length –  What does this tell us? © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Visualiza)on Techniques: Box Plots •  Box Plots –  Invented by J. Tukey –  Another way of displaying the distribu)on of data –  Following figure shows the basic part of a box plot outlier
10th percentile
75th percentile
50th percentile
25th percentile
10th percentile
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Example of Box Plots •  Box plots can be used to compare aFributes © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
What is Data Mining? •  Many Defini.ons –  Non-­‐trivial extrac.on of implicit, previously unknown and poten.ally useful informa.on from data –  Explora.on & analysis, by automa.c or semi-­‐automa.c means, of large quan..es of data in order to discover meaningful paGerns © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
What is (not) Data Mining?   What
is not Data
Mining?
–  Look up phone
number in phone
directory
–  Query a Web
search engine for
information about
“Amazon”
© Tan,Steinbach, Kumar
  What
is Data Mining?
Certain names are more
prevalent in certain US
locations (O’Brien,
O’Rurke, O’Reilly… in
Boston area)
–  Group together similar
documents returned by
search engine according to
their context (e.g. Amazon
rainforest, Amazon.com,)
Introduction to Data Mining
8/05/2005
‹#›
Origins of Data Mining •  Draws ideas from machine learning/AI, paGern recogni.on, sta.s.cs, and database systems •  Tradi.onal Techniques may be unsuitable due to Statistics/ Machine Learning/
–  Enormity of data –  High dimensionality of data –  Heterogeneous, distributed nature of data © Tan,Steinbach, Kumar
Introduction to Data Mining
AI
Pattern
Recognition
Data Mining
Database
systems
8/05/2005
‹#›
Data Mining Tasks •  Predic)on Methods –  Use some variables to predict unknown or future values of other variables. •  Descrip)on Methods –  Find human-­‐interpretable paFerns that describe the data. From [Fayyad, et.al.] Advances in Knowledge Discovery and Data Mining, 1996
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Methods • 
• 
• 
• 
• 
• 
Classifica)on [Predic)ve] Clustering [Descrip)ve] Associa)on Rule Discovery [Descrip)ve] Sequen)al PaFern Discovery [Descrip)ve] Regression [Predic)ve] Anomaly Detec)on [Predic)ve] © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Classifica)on: Defini)on •  Given a collec)on of records (training set ) –  Each record contains a set of a4ributes, one of the aFributes is the class. •  Find a model for class aFribute as a func)on of the values of other aFributes. •  Goal: previously unseen records should be assigned a class as accurately as possible. –  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. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Classifica)on Example Refund Marital
Status
Taxable
Income Cheat
No
No
Single
75K
?
100K
No
Yes
Married
50K
?
Single
70K
No
No
Married
150K
?
Yes
Married
120K
No
Yes
Divorced 90K
?
5
No
Divorced 95K
Yes
No
Single
40K
?
6
No
Married
No
No
Married
80K
?
Tid Refund Marital
Status
Taxable
Income Cheat
1
Yes
Single
125K
2
No
Married
3
No
4
60K
Test
Set
10
7
Yes
Divorced 220K
No
8
No
Single
85K
Yes
9
No
Married
75K
No
10
No
Single
90K
Yes
10
© Tan,Steinbach, Kumar
Training
Set
Introduction to Data Mining
Learn
Classifier
Model
8/05/2005
‹#›
Classifica)on: Applica)on 1 •  Fraud Detec)on –  Goal: Predict fraudulent cases in credit card transac)ons. –  Approach: •  Use credit card transac)ons and the informa)on on its account-­‐
holder as aFributes. –  When does a customer buy, what does he buy, how oden he pays on )me, etc •  Label past transac)ons as fraud or fair transac)ons. This forms the class aFribute. •  Learn a model for the class of the transac)ons. •  Use this model to detect fraud by observing credit card transac)ons on an account. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Classifica)on: Applica)on 2 •  Customer AFri)on/Churn: –  Goal: To predict whether a customer is likely to be lost to a compe)tor. –  Approach: •  Use detailed record of transac)ons with each of the past and present customers, to find aFributes. –  How oden the customer calls, where he calls, what )me-­‐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
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Classifica)on: Applica)on 3 •  Sky Survey Cataloging –  Goal: To predict class (star or galaxy) of sky objects, especially visually faint ones, based on the telescopic survey images (from Palomar Observatory). –  3000 images with 23,040 x 23,040 pixels per image. –  Approach: • 
• 
• 
• 
Segment the image. Measure image aFributes (features) -­‐ 40 of them per object. Model the class based on these features. Success Story: Could find 16 new high red-­‐shid quasars, some of the farthest objects that are difficult to find! From [Fayyad, et.al.] Advances in Knowledge Discovery and Data Mining, 1996
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Classifying Galaxies Courtesy: http://aps.umn.edu
Early
Class:
Attributes:
•  Stages of Formation
•  Image features,
•  Characteristics of light
waves received, etc.
Intermediate
Late
Data Size:
•  72 million stars, 20 million galaxies
•  Object Catalog: 9 GB
•  Image Database: 150 GB
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Clustering Defini)on •  Given a set of data points, each having a set of aFributes, 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 aFributes are con)nuous. –  Other Problem-­‐specific Measures. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Illustra)ng Clustering   Euclidean Distance Based Clustering in 3-D space.
Intracluster distances
are minimized
© Tan,Steinbach, Kumar
Introduction to Data Mining
Intercluster distances
are maximized
8/05/2005
‹#›
Clustering: Applica)on 1 •  Market Segmenta)on: –  Goal: subdivide a market into dis)nct subsets of customers where any subset may conceivably be selected as a market target to be reached with a dis)nct marke)ng mix. –  Approach: •  Collect different aFributes of customers based on their geographical and lifestyle related informa)on. •  Find clusters of similar customers. •  Measure the clustering quality by observing buying paFerns of customers in same cluster vs. those from different clusters. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Clustering: Applica)on 2 •  Document Clustering: –  Goal: To find groups of documents that are similar to each other based on the important terms appearing in them. –  Approach: To iden)fy frequently occurring terms in each document. Form a similarity measure based on the frequencies of different terms. Use it to cluster. –  Gain: Informa)on Retrieval can u)lize the clusters to relate a new document or search term to clustered documents. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Illustra)ng Document Clustering •  Clustering Points: 3204 Ar)cles of Los Angeles Times. •  Similarity Measure: How many words are common in these documents (ader some word filtering). Category
Total
Articles
Correctly
Placed
555
364
Foreign
341
260
National
273
36
Metro
943
746
Sports
738
573
Entertainment
354
278
Financial
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Clustering of S&P 500 Stock Data   Observe Stock Movements every day.
  Clustering points: Stock-{UP/DOWN}
  Similarity Measure: Two points are more similar if the events
described by them frequently happen together on the same day.
v  Association rules can be used to quantify a similarity measure.
Discovered Clusters
1
2
Applied-Matl-DOW N,Bay-Net work-Down,3-COM-DOWN,
Cabletron-Sys-DOWN,CISCO-DOWN,HP-DOWN,
DSC-Co mm-DOW N,INTEL-DOWN,LSI-Logic-DOWN,
Micron-Tech-DOWN,Texas-Inst-Down,Tellabs-Inc-Down,
Natl-Semiconduct-DOWN,Oracl-DOWN,SGI-DOW N,
Sun-DOW N
Apple-Co mp-DOW N,Autodesk-DOWN,DEC-DOWN,
ADV-M icro-Device-DOWN,Andrew-Corp-DOWN,
Co mputer-Assoc-DOWN,Circuit-City-DOWN,
Co mpaq-DOWN, EM C-Corp-DOWN, Gen-Inst-DOWN,
Motorola-DOW N,Microsoft-DOWN,Scientific-Atl-DOWN
3
4
© Tan,Steinbach, Kumar
Fannie-Mae-DOWN,Fed-Ho me-Loan-DOW N,
MBNA-Corp -DOWN,Morgan-Stanley-DOWN
Baker-Hughes-UP,Dresser-Inds-UP,Halliburton-HLD-UP,
Louisiana-Land-UP,Phillips-Petro-UP,Unocal-UP,
Schlu mberger-UP
Introduction to Data Mining
Industry Group
Technology1-DOWN
Technology2-DOWN
Financial-DOWN
Oil-UP
8/05/2005
‹#›
Associa)on Rule Discovery: Defini)on •  Given a set of records each of which contain some number of items from a given collec)on; –  Produce dependency rules which will predict occurrence of an item based on occurrences of other items. Rules Discovered:
TID
Items
1
2
3
4
5
Bread, Coke, Milk
Beer, Bread
Beer, Coke, Diaper, Milk
Beer, Bread, Diaper, Milk
Coke, Diaper, Milk
© Tan,Steinbach, Kumar
Introduction to Data Mining
{Milk} --> {Coke}
{Diaper, Milk} --> {Beer}
8/05/2005
‹#›
Associa)on Rule Discovery: Applica)on 1 •  Marke)ng and Sales Promo)on: –  Let the rule discovered be {Bagels, … } -­‐-­‐> {Potato Chips} –  Potato Chips as consequent => Can be used to determine what should be done to boost its sales. –  Bagels in the antecedent => Can be used to see which products would be affected if the store discon)nues selling bagels. –  Bagels in antecedent and Potato chips in consequent => Can be used to see what products should be sold with Bagels to promote sale of Potato chips! © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Associa)on Rule Discovery: Applica)on 2 •  Supermarket shelf management. –  Goal: To iden)fy items that are bought together by sufficiently many customers. –  Approach: Process the point-­‐of-­‐sale data collected with barcode scanners to find dependencies among items. –  A classic rule -­‐-­‐ •  If a customer buys diaper and milk, then he is very likely to buy beer. •  So, don’t be surprised if you find six-­‐packs stacked next to diapers! © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Associa)on Rule Discovery: Applica)on 3 •  Inventory Management: –  Goal: A consumer appliance repair company wants to an)cipate 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 loca)ons and discover the co-­‐occurrence paFerns. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Sequen)al PaFern Discovery • 
Given is a set of objects, with each object associated with its own 'meline of events, find rules that predict strong sequen)al dependencies among different events. (A B)
• 
(C)
(D E)
Rules are formed by first discovering paFerns. Event occurrences in the paFerns are governed by )ming constraints. (A B)
<= xg
(C)
(D E)
>ng
<= ws
<= ms
© Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Sequen)al PaFern Discovery: Examples •  In telecommunica)ons alarm logs, –  (Inverter_Problem Excessive_Line_Current) (Rec)fier_Alarm) -­‐-­‐> (Fire_Alarm) •  In point-­‐of-­‐sale transac)on sequences, –  Computer Bookstore: (Intro_To_Visual_C) (C++_Primer) -­‐-­‐> (Perl_for_dummies,Tcl_Tk) –  Athle)c Apparel Store: (Shoes) (Racket, Racketball) -­‐-­‐> (Sports_Jacket) © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Regression •  Predict a value of a given con)nuous valued variable based on the values of other variables, assuming a linear or nonlinear model of dependency. •  Greatly studied in sta)s)cs, neural network fields. •  Examples: –  Predic)ng sales amounts of new product based on adve)sing expenditure. –  Predic)ng wind veloci)es as a func)on of temperature, humidity, air pressure, etc. –  Time series predic)on of stock market indices. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Devia)on/Anomaly Detec)on •  Detect significant devia)ons from normal behavior •  Applica)ons: –  Credit Card Fraud Detec)on –  Network Intrusion Detec)on © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Anomaly/Outlier Detec)on •  What are anomalies/outliers? –  The set of data points that are considerably different than the remainder of the data •  Variants of Anomaly/Outlier Detec)on Problems –  Given a database D, find all the data points x ∈ D with anomaly scores greater than some threshold t –  Given a database D, find all the data points x ∈ D having the top-­‐n largest anomaly scores f(x) –  Given a database D, containing mostly normal (but unlabeled) data points, and a test point x, compute the anomaly score of x with respect to D •  Applica)ons: –  Credit card fraud detec)on, telecommunica)on fraud detec)on, network intrusion detec)on, fault detec)on © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Importance of Anomaly Detec)on Ozone Deple)on History • 
In 1985 three researchers (Farman, Gardinar and Shanklin) were puzzled by data gathered by the Bri)sh Antarc)c Survey showing that ozone levels for Antarc)ca had dropped 10% below normal levels • 
Why did the Nimbus 7 satellite, which had instruments aboard for recording ozone levels, not record similarly low ozone concentra)ons? • 
The ozone concentra)ons recorded by the satellite were so low they were being treated as outliers by a computer program and discarded! © Tan,Steinbach, Kumar
Sources:
http://exploringdata.cqu.edu.au/ozone.html
http://www.epa.gov/ozone/science/hole/size.html
Introduction to Data Mining
8/05/2005
‹#›
Anomaly Detec)on Schemes •  General Steps –  Build a profile of the “normal” behavior •  Profile can be paFerns or summary sta)s)cs for the overall popula)on –  Use the “normal” profile to detect anomalies •  Anomalies are observa)ons whose characteris)cs differ significantly from the normal profile •  Types of anomaly detec)on schemes –  Graphical & Sta)s)cal-­‐based –  Distance-­‐based –  Model-­‐based © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Challenges of Data Mining • 
• 
• 
• 
• 
• 
• 
Scalability Dimensionality Complex and Heterogeneous Data Data Quality Data Ownership and Distribu)on Privacy Preserva)on Streaming Data © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›
Next Subject: OLAP •  On-­‐Line Analy)cal Processing (OLAP) was proposed by E. F. Codd, the father of the rela)onal database. •  Rela)onal databases put data into tables, while OLAP uses a mul)dimensional array representa)on. –  Such representa)ons of data previously existed in sta)s)cs and other fields •  There are a number of data analysis and data explora)on opera)ons that are easier with such a data representa)on. © Tan,Steinbach, Kumar
Introduction to Data Mining
8/05/2005
‹#›