Data Mining: Data

Lecture Notes for Chapter 2 Introduction to Data Mining

by Tan, Steinbach, Kumar

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

What is Data?
z

Collection of data objects and their attributes An attribute is a property or characteristic of an object
Examples: eye color of a person, temperature, etc. Attribute is also known as variable, field, characteristic, or feature Objects

Attributes

Tid Refund Marital Status 1 2 3 4 5 6 7 8 9 10

10

Taxable Income Cheat 125K 100K 70K 120K No No No No Yes No No Yes No Yes

Yes No No Yes No No Yes No No No

Single Married Single Married

Divorced 95K Married 60K

A collection of attributes describe an object

Object is also known as record, point, case, sample, entity, or instance

Divorced 220K Single Married Single 85K 75K 90K

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Attribute Values
z

Attribute values are numbers or symbols assigned to an attribute Distinction between attributes and attribute values
Same attribute can be mapped to different attribute values

Example: height can be measured in feet or meters

Different attributes can be mapped pp to the same set of values

Example: Attribute values for ID and age are integers But properties of attribute values can be different

ID has no limit but age has a maximum and minimum value

Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Measurement of Length
z

The way you measure an attribute is somewhat may not match the attributes properties.
5 A B 7 C 8 3 2 1

D 10 4

15

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Types of Attributes
z

There are different types of attributes

Nominal

Examples: ID numbers, numbers eye color, color zip codes Examples: rankings (e.g., taste of potato chips on a scale from 1-10), grades, height in {tall, medium, short} Examples: calendar dates, temperatures in Celsius or Fahrenheit Fahrenheit. Examples: temperature in Kelvin, length, time, counts

Ordinal

Interval

Ratio

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Properties of Attribute Values

z

The type of an attribute depends on which of the following properties it possesses:

Distinctness: Order: Addition: Multiplication: = < > + */

Nominal attribute: distinctness Ordinal attribute: distinctness & order Interval attribute: distinctness, order & addition Ratio attribute: all 4 properties
Introduction to Data Mining 4/18/2004 #

Tan,Steinbach, Kumar

Attribute Type
Nominal

Description
The values of a nominal attribute are just different names, i.e., nominal attributes provide only enough information to distinguish one object from another. (=, ) The values of an ordinal attribute provide enough information to order objects. (<, >)

Examples
zip codes, employee ID numbers, eye color, sex: {male, female}

Operations
mode, entropy, contingency correlation, 2 test

Ordinal

hardness of minerals, {good, better, best}, grades, street numbers

median, percentiles, rank correlation, run tests, sign tests

Interval

For interval attributes, the differences between values are meaningful, i.e., a unit of measurement exists. (+ - ) (+, For ratio variables, both differences and ratios are meaningful. (*, /)

calendar dates, temperature in Celsius or Fahrenheit

mean, standard deviation, Pearson's correlation, t and F tests

Ratio

temperature in Kelvin, monetary quantities, counts, age, mass, length, electrical current

geometric mean, harmonic mean, percent variation

Attribute Level Nominal

Transformation

Comments

Any permutation of values

If all employee ID numbers were reassigned, would it make any difference? An attribute encompassing the notion of good, better best can be represented equally well by the values {1, 2, 3} or by { 0.5, 1, 10}. Thus, the Fahrenheit and Celsius temperature scales differ in terms of where their zero value is and the size of a unit (degree). Length can be measured in meters or feet.

Ordinal

An order preserving change of values, i.e., new_value = f(old_value) where f is a monotonic function.

Interval

new_value =a * old_value + b where a and b are constants

Ratio

new_value = a * old_value

Discrete and Continuous Attributes

z

Discrete Attribute
Has only a finite or countably infinite set of values Examples: zip codes, counts, or the set of words in a collection of d documents t Often represented as integer variables. Note: binary attributes are a special case of discrete attributes

Continuous Attribute
Has real numbers as attribute values Examples: temperature, height, or weight. Practically, real values can only be measured and represented using a finite number of digits. Continuous attributes are typically represented as floating-point variables.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Types of data sets

z

Record
Data Matrix Document Data Transaction Data

Graph
World Wide Web Molecular Structures

Ordered
Spatial Data T Temporal l Data D t Sequential Data Genetic Sequence Data

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Important Characteristics of Structured Data

Dimensionality

Curse of Dimensionality

Sparsity

Only presence counts

Resolution

Patterns depend on the scale

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Record Data
z

Data that consists of a collection of records, each of which consists of a fixed set of attributes
Tid Refund Marital Status 1 2 3 4 5 6 7 8 9 10
10

Taxable Income Cheat 125K 100K 70K 120K No No No No Yes No No Yes No Yes

Yes No No Yes No No Yes No No No

Single Married Single Married

Divorced 95K Married 60K

Divorced 220K Single Married Single 85K 75K 90K

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Data Matrix
z

If data objects have the same fixed set of numeric attributes, then the data objects can be thought of as points in a multi-dimensional space, where each dimension represents a distinct attribute Such data set can be represented by an m by n matrix, where there are m rows, one for each object, and n columns, one for each attribute
Projection of x Load 10.23 12.65 Projection of y load 5.27 6.25 Distance Load Thickness

15.22 16.22
Introduction to Data Mining

2.7 2.2

1.2 1.1
4/18/2004 #

Tan,Steinbach, Kumar

Document Data
z

Each document becomes a `term' vector,

each term is a component (attribute) of the vector, the th value l of f each h component ti is th the number b of f ti times the corresponding term occurs in the document.

timeout

season

coach

game

score

team

ball

lost

pla y

wi n

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Transaction Data
z

A special type of record data, where

each record (transaction) involves a set of items. For F example, l consider id a grocery store. t Th The set t of f products purchased by a customer during one shopping trip constitute a transaction, while the individual products that were purchased are the items.
TID Items

1 2 3 4 5
Tan,Steinbach, Kumar

Bread, Coke, Milk Beer, Bread Beer, Coke, Diaper, Milk Beer, Bread, Diaper, Milk Coke, Diaper, Milk
Introduction to Data Mining 4/18/2004 #

Graph Data
z

Examples: Generic graph and HTML Links

<a a href="papers/papers.html#bbbb"> href papers/papers.html#bbbb Data Mining </a> <li> <a href="papers/papers.html#aaaa"> Graph Partitioning </a> <li> <a href="papers/papers.html#aaaa"> Parallel Solution of Sparse Linear System of Equations </a> <li> <a href="papers/papers.html#ffff"> N-Body Computation and Dense Linear System Solvers

2 5 2 5 1

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Chemical Data
z

Benzene Molecule: C6H6

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Ordered Data
z

Sequences of transactions
Items/Events

An element of the sequence

Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Ordered Data
z

Genomic sequence data

GGTTCCGCCTTCAGCCCCGCGCC CGCAGGGCCCGCCCCGCGCCGTC GAGAAGGGCCCGCCTGGCGGGCG GGGGGAGGCGGGGCCGCCCGAGC CCAACCGAGTCCGACCAGGTGCC CCCTCTGCTCGGCCTAGACCTGA GCTCATTAGGCGGCAGCGGACAG GCCAAGTAGAACACGCGAAGCGC TGGGCTGCCTGCTGCGACCAGGG

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Ordered Data
z

Spatio-Temporal Data

Average Monthly Temperature of land and ocean

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Data Quality
What kinds of data quality problems? z How can we detect problems with the data? z What can we do about these problems?
z

Examples of data quality problems:

Noise and outliers missing i i values l duplicate data

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Noise
z

Noise refers to modification of original values

Examples: distortion of a persons voice when talking on a poor phone and snow snow on television screen

Two Sine Waves

Tan,Steinbach, Kumar Introduction to Data Mining

Two Sine Waves + Noise

4/18/2004 #

Outliers
z

Outliers are data objects with characteristics that are considerably different than most of the other data objects in the data set

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Missing Values
z

Reasons for missing values

Information is not collected (e.g., ( g,p people p decline to g give their age g and weight) g ) Attributes may not be applicable to all cases (e.g., annual income is not applicable to children)

Handling missing values

Eliminate Data Objects Estimate Missing Values Ignore the Missing Value During Analysis Replace with all possible values (weighted by their probabilities)
Introduction to Data Mining 4/18/2004 #

Tan,Steinbach, Kumar

Duplicate Data
z

Data set may include data objects that are duplicates, or almost duplicates of one another
Major issue when merging data from heterogeous sources

Examples:
Same person with multiple email addresses

Data cleaning
Process of dealing with duplicate data issues

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Data Preprocessing
Aggregation z Sampling z Dimensionality Reduction z Feature subset selection z Feature creation z Discretization and Binarization z Attribute Transformation
z

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Aggregation
z

Combining two or more attributes (or objects) into a single attribute (or object) Purpose
Data reduction

Reduce the number of attributes or objects Cities aggregated into regions regions, states states, countries countries, etc Aggregated data tends to have less variability

Change of scale

More stable data

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Aggregation
Variation of Precipitation in Australia

Standard Deviation of Average Monthly Precipitation

Tan,Steinbach, Kumar Introduction to Data Mining

Standard Deviation of Average Yearly Precipitation

4/18/2004 #

Sampling
z

Sampling is the main technique employed for data selection.

It is often used for both the preliminary investigation of the data and the final data analysis.

Statisticians sample because obtaining the entire set of data of interest is too expensive or time consuming. Sampling is used in data mining because processing the entire set of data of interest is too expensive p or time consuming.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Sampling
z

The key principle for effective sampling is the following:

using a sample will work almost as well as using the entire data sets, if the sample is representative A sample is representative if it has approximately the same property (of interest) as the original set of data

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Types of Sampling
z

Simple Random Sampling

There is an equal probability of selecting any particular item

Sampling without replacement

As each item is selected, it is removed from the population

Sampling with replacement

Objects are not removed from the population as they are selected for the sample.
In sampling p g with replacement, p , the same object j can be p picked up p more than once

Stratified sampling
Split the data into several partitions; then draw random samples from each partition

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Sample Size

8000 points

2000 Points

500 Points

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Sample Size
z

What sample size is necessary to get at least one object from each of 10 groups.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Curse of Dimensionality
z

When dimensionality increases, data becomes increasingly sparse in the space that it occupies Definitions of density and distance between points, which is critical for clustering and outlier detection, become less meaningful

Randomly generate 500 points Compute difference between max and min distance between any pair of points

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Dimensionality Reduction
z

Purpose:
Avoid curse of dimensionality Reduce amount of time and memory required by data mining algorithms Allow data to be more easily visualized May help to eliminate irrelevant features or reduce noise

Techniques
Principle Component Analysis Singular Value Decomposition Others: supervised and non-linear techniques

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Dimensionality Reduction: PCA

z

Goal is to find a projection that captures the largest amount of variation in data
x2 e

x1
Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Dimensionality Reduction: PCA

Find the eigenvectors of the covariance matrix z The eigenvectors define the new space
z

x2 e

x1
Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Dimensionality Reduction: ISOMAP

By: Tenenbaum, de Silva, Langford (2000)

z z

Construct a neighbourhood graph For each pair of points in the graph, compute the shortest path distances geodesic distances
Introduction to Data Mining 4/18/2004 #

Tan,Steinbach, Kumar

Dimensionality Reduction: PCA

Dimensions Dimensions= =206 120 160 10 40 80

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Feature Subset Selection

z z

Another way to reduce dimensionality of data Redundant features

duplicate much or all of the information contained in one or more other attributes Example: purchase price of a product and the amount of sales tax paid

Irrelevant features
contain no information that is useful for the data mining task at hand Example: students' ID is often irrelevant to the task of predicting students' GPA

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Feature Subset Selection

z

Techniques:
Brute-force approch:
Try

all possible feature subsets as input to data mining algorithm

Embedded approaches:
Feature selection occurs naturally as part of the data mining algorithm

Filter approaches:

Features are selected before data mining algorithm is run

Wrapper approaches:
Use the data mining algorithm as a black box to find best subset of attributes

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Feature Creation
z

Create new attributes that can capture the important information in a data set much more efficiently than the original attributes Three general methodologies:
Feature Extraction

domain-specific

Mapping Data to New Space Feature Construction

combining features

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Mapping Data to a New Space

z z

Fourier transform Wavelet transform

Two Sine Waves

Two Sine Waves + Noise

Frequency

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Discretization Using Class Labels

z

Entropy based approach

3 categories for both x and y

5 categories for both x and y

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Discretization Without Using Class Labels

Data

Equal interval width

Equal frequency
Tan,Steinbach, Kumar Introduction to Data Mining

K-means
4/18/2004 #

Attribute Transformation
z

A function that maps the entire set of values of a given attribute to a new set of replacement values such that each old value can be identified with one of the new values
Simple functions: xk, log(x), ex, |x| Standardization and Normalization

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Similarity and Dissimilarity

z

Similarity
Numerical measure of how alike two data objects are. Is I higher hi h when h objects bj t are more alike. lik Often falls in the range [0,1]

Dissimilarity
Numerical measure of how different are two data objects Lower when objects are more alike Minimum dissimilarity is often 0 Upper limit varies

Proximity refers to a similarity or dissimilarity

Introduction to Data Mining 4/18/2004 #

Tan,Steinbach, Kumar

Similarity/Dissimilarity for Simple Attributes

p and q are the attribute values for two data objects.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Euclidean Distance
z

Euclidean Distance

dist =

k =1

( pk qk )

Where n is the number of dimensions (attributes) and pk and qk are, respectively, the kth attributes (components) or data objects p and q.
z

Standardization is necessary, if scales differ.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Euclidean Distance

3 2 1
p2 p1 p3 p4

0 0 1 2 3 4 5 6

point p1 p2 p3 p4

x 0 2 3 5

y 2 0 1 1

p1 p p1 p2 p3 p4 0 2.828 3.162 5.099

p2 2.828 0 1.414 3.162

p3 3.162 1.414 0 2

p4 5.099 3.162 2 0

Distance Matrix
Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Minkowski Distance
z

Minkowski Distance is a generalization of Euclidean Distance

dist = ( | pk qk
k =1

1 |r ) r

Where r is a parameter, n is the number of dimensions (attributes) and pk and qk are, respectively, the kth attributes (components) or data objects p and q.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Minkowski Distance: Examples

z

r = 1. City block (Manhattan, taxicab, L1 norm) distance.

A common example of this is the Hamming distance, which is just the number of bits that are different between two binary vectors

z z

r = 2. Euclidean distance r . supremum (Lmax norm, L norm) distance.

This is the maximum difference between any component of the vectors

Do not confuse r with n, i.e., all these distances are defined for all numbers of dimensions.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Minkowski Distance
L1 p1 p2 p3 p4
point p1 p2 p3 p4 x 0 2 3 5 y 2 0 1 1

p1 0 4 4 6 p1 0 2.828 3.162 5.099

p1 p 0 2 3 5

p2 4 0 2 4 p2 2.828 0 1.414 3.162

p2 p 2 0 1 3

p3 4 2 0 2 p3 3.162 1.414 0 2
p3 p 3 1 0 2

p4 6 4 2 0 p4 5.099 3.162 2 0
p4 p 5 3 2 0

L2 p1 p2 p3 p4
L p1 p2 p3 p4

Distance Matrix
Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Mahalanobis Distance

mahalanobi s ( p , q ) = ( p q ) 1 ( p q )T
is i the th covariance i matrix t i of f the input data X

j ,k =

1 n ( X ij X j )( X ik X k ) n 1 i =1

For red points, the Euclidean distance is 14.7, Mahalanobis distance is 6.

Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Mahalanobis Distance
Covariance Matrix:

C B A

0.3 0.2 = 0.2 0.3

A: (0.5, 0.5) B: (0, 1) C: (1.5, 1.5)

Mahal(A,B) = 5 Mahal(A,C) = 4

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Common Properties of a Distance

z

Distances, such as the Euclidean distance, have some well known properties.
1. 2. 3. d(p, q) 0 for all p and q and d(p, q) = 0 only if p = q. (Positive definiteness) d(p, q) = d(q, p) for all p and q. (Symmetry) d(p, r) d(p, q) + d(q, r) for all points p, q, and r. (Triangle Inequality)

where d(p, q) is the distance (dissimilarity) between points (data objects), objects) p and q.
z

A distance that satisfies these properties is a metric

Introduction to Data Mining 4/18/2004 #

Tan,Steinbach, Kumar

Common Properties of a Similarity

z

Similarities, also have some well known properties.

1. 2. s(p, q) = 1 (or maximum similarity) only if p = q. s(p, q) = s(q, p) for all p and q. (Symmetry)

where s(p, q) is the similarity between points (data objects), p and q.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Similarity Between Binary Vectors

z

Common situation is that objects, p and q, have only binary attributes Compute p similarities using g the following gq quantities
M01 =thenumberofattributeswherepwas0andqwas1 M10=thenumberofattributeswherepwas1andqwas0 M00 =thenumberofattributeswherepwas0andqwas0 M11 =thenumberofattributeswherepwas1andqwas1

Simple Matching and Jaccard Coefficients

SMC = number of matches / number of attributes = (M11 + M00) / (M01 + M10 + M11 + M00)

J = number of 11 matches / number of not-both-zero attributes values = (M11) / (M01 + M10 + M11)

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

SMC versus Jaccard: Example

p= 1000000000 q= 0000001001
M01 =2(thenumberofattributeswherepwas0andqwas1) M10 =1(thenumberofattributeswherepwas1andqwas0) M00 =7(thenumberofattributeswherepwas0andqwas0) M11 =0(thenumberofattributeswherepwas1andqwas1)

SMC = (M11 + M00)/(M01 + M10 + M11 + M00) = (0+7) / (2+1+0+7) = 0 0.7 7

J = (M11) / (M01 + M10 + M11) = 0 / (2 + 1 + 0) = 0

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Cosine Similarity
z

If d1 and d2 are two document vectors, then cos( d1, d2 ) = (d1 d2) / ||d1|| ||d2|| ,
where indicates vector dot product and || d || is the length of vector d.

Example:
d1 = 3 2 0 5 0 0 0 2 0 0 d2 = 1 0 0 0 0 0 0 1 0 2
d1 d2= 3*1 + 2*0 + 0*0 + 5*0 + 0*0 + 0*0 + 0*0 + 2*1 + 0*0 + 0*2 = 5 ||d1|| = (3*3+2*2+0*0+5*5+0*0+0*0+0*0+2*2+0*0+0*0)0.5 = (42) 0.5 = 6.481 ||d2|| = (1*1+0*0+0*0+0*0+0*0+0*0+0*0+1*1+0*0+2*2) 0.5 = (6) 0.5 = 2.245

cos( d1, d2 ) = .3150

Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Extended Jaccard Coefficient (Tanimoto)

z

Variation of Jaccard for continuous or count attributes

Reduces to Jaccard for binary attributes

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Correlation
Correlation measures the linear relationship between objects z To T compute t correlation, l ti we standardize t d di d data t objects, p and q, and then take their dot product
z

= ( pk mean( p)) / std ( p) pk

= ( qk mean( q)) / std qk td ( q)

correlation( p, q) = p q
Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 #

Visually Evaluating Correlation

Scatter plots showing the similarity from 1 to 1.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

General Approach for Combining Similarities

z

Sometimes attributes are of many different types, but an overall similarity is needed.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Using Weights to Combine Similarities

z

May not want to treat all attributes the same.

Use weights wk which are between 0 and 1 and sum to 1 1.

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Density
z

Density-based clustering require a notion of density Examples:

Euclidean density

Euclidean density = number of points per unit volume

Probability density Graph-based density

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Euclidean Density Cell-based

z

Simplest approach is to divide region into a number of rectangular cells of equal volume and define density as # of points the cell contains

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004

Euclidean Density Center-based

z

Euclidean density is the number of points within a specified radius of the point

Tan,Steinbach, Kumar

Introduction to Data Mining

4/18/2004