Based on slides from Han J., et. al.

2013

1
Chapter 2: Getting to Know Your
Data
Data Objects and Attribute Types

Basic Statistical Descriptions of Data

Data Visualization

Measuring Data Similarity and Dissimilarity

Summary

2
Types of Data Sets
 Record
 Relational records
 Data matrix, e.g., numerical matrix, crosstabs

timeout

season
coach

game
score
team

ball

lost
pla

wi
n
y
 Document data: text documents: term-
frequency vector
 Transaction data
Document 1 3 0 5 0 2 6 0 2 0 2
 Graph and network
 Document 2 0 7 0 2 1 0 0 3 0 0
World Wide Web
 Social or information networks Document 3 0 1 0 0 1 2 2 0 3 0
 Molecular Structures
 Ordered
 Video data: sequence of images TID Items
 Temporal data: time-series 1 Bread, Coke, Milk
 Sequential Data: transaction sequences 2 Beer, Bread
 Genetic sequence data 3 Beer, Coke, Diaper, Milk
 Spatial, image and multimedia:
4 Beer, Bread, Diaper, Milk
 Spatial data: maps

5 Coke, Diaper, Milk
Image data:
 Video data:

3
Important Characteristics of Structured Data

Dimensionality
 Curse of dimensionality

Sparsity
 Only presence counts

Resolution
 Patterns depend on the scale

Distribution
 Centrality and dispersion

4
Data Objects

Data sets are made up of data objects.

A data object represents an entity.
Examples:
 sales database: customers, store items, sales
 medical database: patients, treatments
 university database: students, professors, courses
Also called samples , examples, instances, data points, objects,
tuples.
Data objects are described by attributes.
Database rows -> data objects; columns ->attributes.

5
Attributes
Attribute (or dimensions, features, variables): a data field,
representing a characteristic or feature of a data object.
E.g., customer _ID, name, address
Types:
Nominal
Binary
Numeric: quantitative
Interval-scaled

Ratio-scaled

6
Attribute Types
 Nominal: categories, states, or “names of things”
 Hair_color = {auburn, black, blond, brown, grey, red, white}
 marital status, occupation, ID numbers, zip codes
 Binary
 Nominal attribute with only 2 states (0 and 1)
 Symmetric binary: both outcomes equally important
 e.g., gender
 Asymmetric binary: outcomes not equally important.
 e.g., medical test (positive vs. negative)
 Convention: assign 1 to most important outcome (e.g., HIV positive)

 Ordinal
 Values have a meaningful order (ranking) but magnitude between
successive values is not known.
 Size = {small, medium, large}, grades, army rankings

7
Numeric Attribute Types
Quantity (integer or real-valued)
Interval
 Measured on a scale of equal-sized units
 Values have order
 E.g., temperature in C˚or F˚, calendar dates
 No true zero-point
Ratio
 Inherent zero-point
 We can speak of values as being an order of magnitude larger
than the unit of measurement (10 K˚ is twice as high as 5 K˚).
 e.g., temperature in Kelvin, length, counts, monetary
quantities

8
Discrete vs. Continuous Attributes
Discrete Attribute
Has only a finite or countably infinite set of values
 E.g., zip codes, profession, or the set of words in a collection of
documents
Sometimes, represented as integer variables
Note: Binary attributes are a special case of discrete attributes
Continuous Attribute
Has real numbers as attribute values
 E.g., 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

9
Chapter 2: Getting to Know Your
Data
Data Objects and Attribute Types

Basic Statistical Descriptions of Data

Data Visualization

Measuring Data Similarity and Dissimilarity

Summary

10
Basic Statistical Descriptions of Data
 Motivation
To better understand the data: central tendency,
variation and spread
 Data dispersion characteristics
median, max, min, quantiles, outliers, variance, etc.
 Numerical dimensions correspond to sorted intervals
Data dispersion: analyzed with multiple
granularities of precision
Boxplot or quantile analysis on sorted intervals
 Dispersion analysis on computed measures
Folding measures into numerical dimensions
Boxplot or quantile analysis on the transformed cube

11
Measuring the Central Tendency
 Mean (algebraic measure) (sample vs. population): 1 n
x   xi   x
Note: n is sample size and N is population size. n i 1 N
n
 Weighted arithmetic mean:

w x i i
Trimmed mean: chopping extreme values x  i 1n
 Median: w
i 1
i
 Middle value if odd number of values, or average of the middle
two values otherwise
 Estimated by interpolation (for grouped data):
n / 2  ( freq ) l Media


median L1  ( ) width n
Mode freq median interv
al

 Value that occurs most frequently in the data

 Unimodal, bimodal, trimodal
 Empirical formula:
mean  mode 3 (mean  median)
12
 Symmetric vs. Skewed Data
Median, mean and mode of symmetric, symmetric

positively and negatively skewed data

positively skewed negatively skewed

Data Mining: Concepts and 13

April 20, 2025Techniques
Measuring the Dispersion of Data
 Quartiles, outliers and boxplots
 Quartiles: Q1 (25th percentile), Q3 (75th percentile)

 Inter-quartile range: IQR = Q3 – Q1

 Five number summary: min, Q1, median, Q3, max

 Boxplot: ends of the box are the quartiles; median is marked; add whiskers, and

plot outliers individually

 Outlier: usually, a value higher/lower than 1.5 x IQR

 Variance and standard deviation (sample: s, population: σ)

 Variance: (algebraic, scalable computation)
1 n 1 n 2 1 n
s  2
 i
n  1 i 1
( x  x ) 2
 [ i n
n  1 i 1
x  (
i 1
xi ]
) 2

 Standard deviation s (or σ) is the square

n root ofn variance s2 (or σ2)
1 1
 x
2
2  ( xi   ) 2  i  2
N i 1 N i 1 14
Boxplot Analysis
 Five-number summary of a distribution
 Minimum, Q1, Median, Q3, Maximum

 Boxplot
 Data is represented with a box
 The ends of the box are at the first and third
quartiles, i.e., the height of the box is IQR
 The median is marked by a line within the box
 Whiskers: two lines outside the box extended to
Minimum and Maximum
 Outliers: points beyond a specified outlier
threshold, plotted individually

15
Visualization of Data Dispersion: 3-D Boxplots

Data Mining: Concepts and 16

April 20, 2025 Techniques
Properties of Normal Distribution Curve

The normal (distribution) curve

From μ–σ to μ+σ: contains about 68% of the measurements (μ:
mean, σ: standard deviation)
 From μ–2σ to μ+2σ: contains about 95% of it
From μ–3σ to μ+3σ: contains about 99.7% of it

17
Graphic Displays of Basic Statistical Descriptions

 Boxplot: graphic display of five-number summary

 Histogram: x-axis are values, y-axis repres. frequencies
 Quantile plot: each value xi is paired with fi indicating that
approximately 100 fi % of data are  xi
 Quantile-quantile (q-q) plot: graphs the quantiles of one univariant
distribution against the corresponding quantiles of another
 Scatter plot: each pair of values is a pair of coordinates and plotted
as points in the plane

18
Histogram Analysis
Histogram: Graph display of tabulated
40
frequencies, shown as bars
It shows what proportion of cases fall 35
into each of several categories 30

Differs from a bar chart in that it is the25

area of the bar that denotes the value, 20
not the height as in bar charts, a crucial15
distinction when the categories are not10
of uniform width
5
The categories are usually specified as
0
non-overlapping intervals of some 10000 30000 50000 70000 90000

variable. The categories (bars) must be

adjacent
19
Histograms Often Tell More than Boxplots

 The two histograms

shown in the left may
have the same boxplot
representation
 The same values for:
min, Q1, median, Q3,
max
 But they have rather
different data
distributions

20
Quantile Plot
Displays all of the data (allowing the user to assess both the
overall behavior and unusual occurrences)
Plots quantile information
For a data x data sorted in increasing order, f indicates that
i i
approximately 100 fi% of the data are below or equal to the
value xi

Data Mining: Concepts and 21

Techniques
Quantile-Quantile (Q-Q) Plot
Graphs the quantiles of one univariate distribution against the
corresponding quantiles of another
View: Is there is a shift in going from one distribution to another?
Example shows unit price of items sold at Branch 1 vs. Branch 2 for
each quantile. Unit prices of items sold at Branch 1 tend to be lower
than those at Branch 2.

22
Scatter plot
Provides a first look at bivariate data to see clusters of points,
outliers, etc
Each pair of values is treated as a pair of coordinates and plotted
as points in the plane

23
Positively and Negatively Correlated Data

 The left half fragment is positively

correlated
 The right half is negative correlated

24
Uncorrelated Data

25
Chapter 2: Getting to Know Your Data

Data Objects and Attribute Types

Basic Statistical Descriptions of Data

Data Visualization

Measuring Data Similarity and Dissimilarity

Summary

26
Data Visualization
 Why data visualization?
 Gain insight into an information space by mapping data onto graphical
primitives
 Provide qualitative overview of large data sets
 Search for patterns, trends, structure, irregularities, relationships among data
 Help find interesting regions and suitable parameters for further quantitative
analysis
 Provide a visual proof of computer representations derived
 Categorization of visualization methods:
 Pixel-oriented visualization techniques
 Geometric projection visualization techniques
 Icon-based visualization techniques
 Hierarchical visualization techniques
 Visualizing complex data and relations

27
 Pixel-Oriented Visualization Techniques
 For a data set of m dimensions, create m windows on the screen, one for
each dimension
 The m dimension values of a record are mapped to m pixels at the
corresponding positions in the windows
 The colors of the pixels reflect the corresponding values

(a) Income (b) Credit (c) transaction (d) age

Limit volume 28
Laying Out Pixels in Circle Segments
 To save space and show the connections among multiple dimensions, space
filling is often done in a circle segment

(a) Representing a data

(b) Laying out pixels in circle
record in circle segment
Representing about 265,000 50-dimensional Data Items segment
with the ‘Circle Segments’ Technique 29
Geometric Projection Visualization Techniques
Visualization of geometric transformations and projections of the
data
Methods
Direct visualization
Scatterplot and scatterplot matrices
Landscapes
Projection pursuit technique: Help users find meaningful
projections of multidimensional data
Prosection views
Hyperslice
Parallel coordinates
30
 31
Data Mining: Concepts and
Techniques
Ribbons with Twists Based on Vorticity
Scatterplot Matrices
Used by ermission of M. Ward, Worcester Polytechnic Institute

Matrix of scatterplots (x-y-diagrams) of the k-dim. data [total of (k2/2-k) scatterplots]

32
Landscapes
Used by permission of B. Wright, Visible Decisions Inc.

news articles
visualized as
a landscape

 Visualization of the data as perspective landscape

 The data needs to be transformed into a (possibly artificial) 2D spatial
representation which preserves the characteristics of the data
33
 Parallel Coordinates
 n equidistant axes which are parallel to one of the screen axes and correspond
to the attributes
 The axes are scaled to the [minimum, maximum]: range of the corresponding
attribute
 Every data item corresponds to a polygonal line which intersects each of the
axes at the point which corresponds to the value for the attribute

34
Parallel Coordinates of a Data Set

35
Icon-Based Visualization Techniques
Visualization of the data values as features of icons
Typical visualization methods
Chernoff Faces
Stick Figures
General techniques
Shape coding: Use shape to represent certain information
encoding
Color icons: Use color icons to encode more information
Tile bars: Use small icons to represent the relevant feature
vectors in document retrieval

36
Chernoff Faces
 A way to display variables on a two-dimensional surface, e.g., let x be
eyebrow slant, y be eye size, z be nose length, etc.
 The figure shows faces produced using 10 characteristics--head eccentricity,
eye size, eye spacing, eye eccentricity, pupil size, eyebrow slant, nose size,
mouth shape, mouth size, and mouth opening): Each assigned one of 10
possible values, generated using Mathematica (S. Dickson)

 REFERENCE: Gonick, L. and Smith, W.

The Cartoon Guide to Statistics. New York:
Harper Perennial, p. 212, 1993
 Weisstein, Eric W. "Chernoff Face." From
MathWorld--A Wolfram Web Resource.
mathworld.wolfram.com/ChernoffFace.html

37
 Stick Figure
A census data
figure showing
age, income,
used by permission of G. Grinstein, University of Massachusettes at Lowell

gender,
education, etc.

A 5-piece stick
figure (1 body
and 4 limbs w.
different
angle/length)

Data Mining: Concepts and 38

Techniques
Hierarchical Visualization Techniques
Visualization of the data using a hierarchical partitioning
into subspaces
Methods
Dimensional Stacking
Worlds-within-Worlds
Tree-Map
Cone Trees
InfoCube

39
 Dimensional Stacking

Partitioning of the n-dimensional attribute space in 2-D subspaces,

which are ‘stacked’ into each other
Partitioning of the attribute value ranges into classes. The
important attributes should be used on the outer levels.
Adequate for data with ordinal attributes of low cardinality
But, difficult to display more than nine dimensions
Important to map dimensions appropriately

40
Dimensional Stacking
Used by permission of M. Ward, Worcester Polytechnic Institute

Visualization of oil mining data with longitude and latitude mapped to the
outer x-, y-axes and ore grade and depth mapped to the inner x-, y-axes
41
 Worlds-within-Worlds
 Assign the function and two most important parameters to innermost world
 Fix all other parameters at constant values - draw other (1 or 2 or 3
dimensional worlds choosing these as the axes)
 Software that uses this paradigm

 N–vision: Dynamic
interaction through data
glove and stereo displays,
including rotation, scaling
(inner) and translation
(inner/outer)
 Auto Visual: Static
interaction by means of
queries
42
 Tree-Map
Screen-filling method which uses a hierarchical partitioning of the
screen into regions depending on the attribute values
The x- and y-dimension of the screen are partitioned alternately
according to the attribute values (classes)

Schneiderman@UMD: Tree-Map of a File System Schneiderman@UMD: Tree-Map to

support large data sets of a million
43
items
InfoCube
A 3-D visualization technique where hierarchical information
is displayed as nested semi-transparent cubes
The outermost cubes correspond to the top level data, while
the subnodes or the lower level data are represented as
smaller cubes inside the outermost cubes, and so on

44
 Three-D Cone Trees
 3D cone tree visualization technique works well
for up to a thousand nodes or so
 First build a 2D circle tree that arranges its
nodes in concentric circles centered on the root
node
 Cannot avoid overlaps when projected to 2D
 G. Robertson, J. Mackinlay, S. Card. “Cone
Trees: Animated 3D Visualizations of
Hierarchical Information”, ACM SIGCHI'91
 Graph from Nadeau Software Consulting
website: Visualize a social network data set that
models the way an infection spreads from one
person to the next

45
Visualizing Complex Data and Relations
Visualizing non-numerical data: text and social networks
Tag cloud: visualizing user-generated tags
 The importance of tag is
represented by font
size/color
 Besides text data, there are
also methods to visualize
relationships, such as
visualizing social networks

Newsmap: Google News Stories in 2005

Chapter 2: Getting to Know Your Data

Data Objects and Attribute Types

Basic Statistical Descriptions of Data

Data Visualization

Measuring Data Similarity and Dissimilarity

Summary

47
Similarity and Dissimilarity
Similarity
Numerical measure of how alike two data objects are
Value is higher when objects are more alike
Often falls in the range [0,1]

Dissimilarity (e.g., distance)

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

Proximity refers to a similarity or dissimilarity

48
Data Matrix and Dissimilarity Matrix
Data matrix
n data points with p  x11 ... x1f ... x1p 
 
dimensions  ... ... ... ... ... 
Two modes x ... xif ... xip 
 i1 
 ... ... ... ... ... 
x ... xnf ... xnp 
 n1 
Dissimilarity matrix
n data points, but  0 
 d(2,1) 0 
registers only the  
 d(3,1) d ( 3,2) 0 
distance  
A triangular matrix  : : : 
 d ( n,1) d ( n,2) ... ... 0
Single mode
49
Proximity Measure for Nominal Attributes

Can take 2 or more states, e.g., red, yellow, blue, green

(generalization of a binary attribute)
Method 1: Simple matching
m: # of matches, p: total # of variables

d (i, j)  p p m
Method 2: Use a large number of binary attributes
creating a new binary attribute for each of the M nominal
states

50
Proximity Measure for Binary Attributes
Object j
 A contingency table for binary data
Object i

 Distance measure for symmetric

binary variables:
 Distance measure for asymmetric
binary variables:
 Jaccard coefficient (similarity measure
for asymmetric binary variables):

 Note: Jaccard coefficient is the same as “coherence”:

51
Dissimilarity between Binary Variables
Example
Name Gender Fever Cough Test-1 Test-2 Test-3 Test-4
Jack M Y N P N N N
Mary F Y N P N P N
Jim M Y P N N N N

 Gender is a symmetric attribute

 The remaining attributes are asymmetric binary
 Let the values Y and P be 1, and the value N 0

0 1
d ( jack , mary )  0.33
2  0 1
11
d ( jack , jim )  0.67
111
1 2
d ( jim , mary )  0.75
11 2
52
Standardizing Numeric Data
 Z-score:
x
z  
 X: raw score to be standardized, μ: mean of the population, σ: standard
deviation
 the distance between the raw score and the population mean in units of
the standard deviation
 negative when the raw score is below the mean, “+” when above
 An alternative way: Calculate the mean absolute deviation
s f 1n (| x1 f  m f |  | x2 f  m f | ... | xnf  m f |)
where
m f  1n (x1 f  x2 f  ...  xnf )
xif  m f
.

 standardized measure (z-score):

zif  sf
 Using mean absolute deviation is more robust than using standard deviation

53
Example: Data Matrix and Dissimilarity Matrix
Data Matrix
x2 x4
point attribute1 attribute2
4 x1 1 2
x2 3 5
x3 2 0
x4 4 5
2 x1
Dissimilarity Matrix
(with Euclidean Distance)
x3
0 4 x1 x2 x3 x4
2
x1 0
x2 3.61 0
x3 2.24 5.1 0
x4 4.24 1 5.39 0
54
Distance on Numeric Data: Minkowski Distance
 Minkowski distance: A popular distance measure

where i = (xi1, xi2, …, xip) and j = (xj1, xj2, …, xjp) are two p-
dimensional data objects, and h is the order (the distance so
defined is also called L-h norm)
 Properties
 d(i, j) > 0 if i ≠ j, and d(i, i) = 0 (Positive definiteness)

 d(i, j) = d(j, i) (Symmetry)

 d(i, j)  d(i, k) + d(k, j) (Triangle Inequality)

 A distance that satisfies these properties is a metric

55
 Special Cases of Minkowski Distance
 h = 1: Manhattan (city block, L1 norm) distance
 E.g., the Hamming distance: the number of bits that are different
between two binary vectors
d (i, j) | x  x |  | x  x | ... | x  x |
i1 j1 i2 j 2 ip jp

 h = 2: (L2 norm) Euclidean distance

d (i, j)  (| x  x |2  | x  x |2 ... | x  x |2 )
i1 j1 i2 j 2 ip jp

 h  . “supremum” (Lmax norm, L norm) distance.

 This is the maximum difference between any component (attribute) of
the vectors

56
 Example: Minkowski Distance
Dissimilarity Matrices
point attribute 1 attribute 2 Manhattan
x1 1 2 (L1)L x1 x2 x3 x4
x2 3 5 x1 0
x3 2 0 x2 5 0
x4 4 5 x3 3 6 0
x4 6 1 7 0
Euclidean (L2)
x2 x4
L2 x1 x2 x3 x4
4 x1 0
x2 3.61 0
x3 2.24 5.1 0
x4 4.24 1 5.39 0

2 x1
Supremum
L x1 x2 x3 x4
x1 0
x2 3 0
x3 x3 2 5 0
0 2 4 x4 3 1 5 0
57
 Ordinal Variables
An ordinal variable can be discrete or continuous
Order is important, e.g., rank
Can be treated like interval-scaled
replace xif by their rank rif {1,..., M f }
map the range of each variable onto [0, 1] by replacing i-th
object in the f-th variable by
rif  1
zif 
Mf 1

compute the dissimilarity using methods for interval-scaled

variables
58
Attributes of Mixed Type
A database may contain all attribute types
Nominal, symmetric binary, asymmetric binary, numeric,
ordinal
One may use a weighted formula to combine their effects
 pf 1 ij( f ) dij( f )
d (i, j)  p
 f 1 ij( f )
f is binary or nominal:
dij(f) = 0 if xif = xjf , or dij(f) = 1 otherwise
f is numeric: use the normalized distance
f is ordinal
 Compute ranks rif and
zif  r
if1
 Treat z as interval-scaled
if M 1 f

59
 Cosine Similarity
 A document can be represented by thousands of attributes, each recording the
frequency of a particular word (such as keywords) or phrase in the document.

 Other vector objects: gene features in micro-arrays, …

 Applications: information retrieval, biologic taxonomy, gene feature mapping, ...
 Cosine measure: If d and d are two vectors (e.g., term-frequency vectors), then
1 2

cos(d1, d2) = (d1  d2) /||d1|| ||d2|| ,

where  indicates vector dot product, ||d||: the length of vector d

60
 Example: Cosine Similarity
 cos(d , d ) = (d  d ) /||d || ||d || ,
1 2 1 2 1 2
where  indicates vector dot product, ||d|: the length of vector d

 Ex: Find the similarity between documents 1 and 2.

d1 = (5, 0, 3, 0, 2, 0, 0, 2, 0, 0)
d2 = (3, 0, 2, 0, 1, 1, 0, 1, 0, 1)

d1d2 = 5*3+0*0+3*2+0*0+2*1+0*1+0*1+2*1+0*0+0*1 = 25
||d1||= (5*5+0*0+3*3+0*0+2*2+0*0+0*0+2*2+0*0+0*0)0.5=(42)0.5 = 6.481
||d2||= (3*3+0*0+2*2+0*0+1*1+1*1+0*0+1*1+0*0+1*1)0.5=(17)0.5 = 4.12
cos(d1, d2 ) = 0.94

61
 KL Divergence: Comparing Two
Probability Distributions
 The Kullback-Leibler (KL) divergence: Measure the difference between two
probability distributions over the same variable x
 From information theory, closely related to relative entropy, information
divergence, and information for discrimination
 DKL(p(x) || q(x)): divergence of q(x) from p(x), measuring the information lost
when q(x) is used to approximate p(x)
 Discrete form:

 The KL divergence measures the expected number of extra bits required to code
samples from p(x) (“true” distribution) when using a code based on q(x), which
represents a theory, model, description, or approximation of p(x)
 Its continuous form:

 The KL divergence: not a distance measure, not a metric: asymmetric, not

satisfy triangular inequality
62
How to Compute the KL
Divergence?
 Base on the formula, DKL(P,Q) ≥ 0 and DKL(P || Q) = 0 if and only if P = Q.
 How about when p = 0 or q = 0?
 limp→0 p log p = 0
 when p != 0 but q = 0, DKL(p || q) is defined as ∞, i.e., if one event e is possible
(i.e., p(e) > 0), and the other predicts it is absolutely impossible (i.e., q(e) = 0),
then the two distributions are absolutely different
 However, in practice, P and Q are derived from frequency distributions, not
counting the possibility of unseen events. Thus smoothing is needed
 Example: P : (a : 3/5, b : 1/5, c : 1/5). Q : (a : 5/9, b : 3/9, d : 1/9)
 need to introduce a small constant ϵ, e.g., ϵ = 10−3
 The sample set observed in P, SP = {a, b, c}, SQ = {a, b, d}, SU = {a, b, c, d}
 Smoothing, add missing symbols to each distribution, with probability ϵ
 P′ : (a : 3/5 − ϵ/3, b : 1/5 − ϵ/3, c : 1/5 − ϵ/3, d : ϵ)
 Q′ : (a : 5/9 − ϵ/3, b : 3/9 − ϵ/3, c : ϵ, d : 1/9 − ϵ/3).
 DKL(P’ || Q’) can be computed easily
63
Chapter 2: Getting to Know Your Data

Data Objects and Attribute Types

Basic Statistical Descriptions of Data

Data Visualization

Measuring Data Similarity and Dissimilarity

Summary

64
Summary
Data attribute types: nominal, binary, ordinal, interval-scaled,
ratio-scaled
Many types of data sets, e.g., numerical, text, graph, Web, image.
Gain insight into the data by:
Basic statistical data description: central tendency, dispersion,
graphical displays
Data visualization: map data onto graphical primitives
Measure data similarity
Above steps are the beginning of data preprocessing
Many methods have been developed but still an active area of
research
References
 W. Cleveland, Visualizing Data, Hobart Press, 1993
 T. Dasu and T. Johnson. Exploratory Data Mining and Data Cleaning. John Wiley,
2003
 U. Fayyad, G. Grinstein, and A. Wierse. Information Visualization in Data Mining and
Knowledge Discovery, Morgan Kaufmann, 2001
 L. Kaufman and P. J. Rousseeuw. Finding Groups in Data: an Introduction to Cluster
Analysis. John Wiley & Sons, 1990.
 H. V. Jagadish et al., Special Issue on Data Reduction Techniques. Bulletin of the
Tech. Committee on Data Eng., 20(4), Dec. 1997
 D. A. Keim. Information visualization and visual data mining, IEEE trans. on
Visualization and Computer Graphics, 8(1), 2002
 D. Pyle. Data Preparation for Data Mining. Morgan Kaufmann, 1999
 S. Santini and R. Jain,” Similarity measures”, IEEE Trans. on Pattern Analysis and
Machine Intelligence, 21(9), 1999
 E. R. Tufte. The Visual Display of Quantitative Information, 2 nd ed., Graphics Press,
2001
 C. Yu et al., Visual data mining of multimedia data for social and behavioral studies,
Information Visualization, 8(1), 2009