Data mining

CLASSIFICATION
‫کننده‬ ‫تنظیم‬:‫اسدی‬ ‫میثم‬ meysam_tabriz@yahoo.com

GENERAL APPROACH TO CLASSIFICATION:
1. decision tree classifiers
2. Bayesian classifiers
3. Rule-based classifiers
4. Classification Using Frequent Patterns
5. Lazy Learners
 k-Nearest-Neighbor Classifiers(KNN)
 Case-Based Reasoning (CBR)
6. Neural Networks (ANN)
 Classification by Backpropagation
 multilayer perceptron (MLP)
 radial basis function ( RBF)
 Support Vector Machines (SVM)
 learning vector quantization ( LVQ)
7. Other Classification Method
 Genetic Algorithms
 Rough Set Approach
 Fuzzy Set Approaches

1. DECISION TREE CLASSIFIERS:
ID3 (ITERATIVE DICHOTOMISER)
C4.5 ( CONCEPT LEARNING SYSTEMS)
CART (CLASSIFICATION AND REGRESSION TREES)
CHAID
C-SEP
G-STATISTIC

2. BAYESIAN CLASSIFIERS

3. RULE-BASED CLASSIFIERS:
USING IF-THEN RULES FOR CLASSIFICATION
RULE EXTRACTION FROM A DECISION TREE
RULE INDUCTION USING A SEQUENTIAL COVERING ALGORITHM
REP
I-REP
FOIL
AQ
CN2
RIPPER

ARTIFICIAL NEURAL NETWORKS (ANN)

4. CLASSIFICATION BY BACKPROPAGATION:

5. SUPPORT VECTOR MACHINES (SVM)

6. CLASSIFICATION USING FREQUENT
PATTERNS

7. LAZY LEARNERS:
K-NEAREST-NEIGHBOR CLASSIFIERS(KNN)
CASE-BASED REASONING (CBR)

•K-NEAREST-NEIGHBOR CLASSIFIERS(KNN)

•CASE-BASED REASONING (CBR)

8. OTHER CLASSIFICATION METHOD GENETIC
GENETIC ALGORITHMS
ROUGH SET APPROACH
FUZZY SET APPROACHES
NEURAL NETWORKS (ANN): MLP-RBF-SVM-LVQ

GENETIC ALGORITHMS

FUZZY SET APPROACHES

ROUGH SET APPROACH

CLUSTERING

OVERVIEW OF BASIC CLUSTERING METHODS:
.( partitioning methods) :
k-Means
k-Medoids
PAM(–CLARA–CLARANS)
.hierarchical methods)) :
DIANA
AGNES
BIRCH
Chameleon
CURE
ROCK

3.‫ل‬ ‫چگ‬ ‫ی‬ ‫کم‬:density-based methods))
•‫م‬ ‫ی‬ ‫لگ‬DBSCAN
•‫م‬ ‫ی‬ ‫لگ‬OPTICS
•‫م‬ ‫ی‬ ‫لگ‬DENCLUE
4.‫ی‬ ‫گ‬ ‫ج‬ ‫ط‬ ‫ک‬:(grid-based methods)
‫م‬ ‫ی‬ ‫لگ‬STING
‫م‬ ‫ی‬ ‫لگ‬CLIQUE
5.:(Probabilistic model-based clustering)
‫م‬ ‫ی‬ ‫لگ‬EM
6.‫ع‬ ‫ج‬ ‫م‬ ‫ی‬
‫م‬ ‫ی‬ ‫لگ‬FCM
7.‫ق‬ ‫عص‬ ‫ک‬:‫ن‬ ‫ت‬ ‫گ‬(SOM:Self-Organizing map)

( PARTITIONING METHODS) :

KMEANS

KMEDOIDS

‫تکنیک‬ ‫دیگر‬ ‫های‬ ‫اگوریتم‬( PARTITIONING METHODS)
1.PAM : PARTITIONING AROUND MEDOIDS
2.CLARA : CLUSTERING LARGE APPLICATIONS
3.CLARANS : CLUSTERING LARGE APPLICATIONS BASED UPON RANDOMIZED SEARCH

PAM : PARTITIONING AROUND MEDOIDS

CLARA : CLUSTERING LARGE APPLICATIONS

3.CLARANS : CLUSTERING LARGE APPLICATIONS BASED UPON RANDOMIZED SEARCH

:HIERARCHICAL METHODS))
‫م‬ ‫ی‬ ‫لگ‬DIANA
‫م‬ ‫ی‬ ‫لگ‬AGNES
‫م‬ ‫ی‬ ‫لگ‬BIRCH
‫م‬ ‫ی‬ ‫لگ‬Chameleon
‫م‬ ‫ی‬ ‫لگ‬CURE
•‫م‬ ‫ی‬ ‫لگ‬ROCK

‫م‬ ‫ی‬ ‫لگ‬DIANA & AGNES

BIRCH(BALANCED ITERATIVE REDUCING AND CLUSTERING USING HIERARCHIE )

‫م‬ ‫ی‬ ‫لگ‬CURE (CLUSTERING USING REPRESENTATIVES)

(ROBUST CLUSTERING USING LINKS)ROCK

‫الگوریتم‬CHAMELEON
Chameleon: hierarchical
clustering based on k-nearest
neighbors and dynamic modeling.

‫داده‬ ‫چگالی‬ ‫یا‬ ‫تراکم‬ ‫بر‬ ‫مبتنی‬ ‫بندی‬ ‫خوشه‬:DENSITY-BASED METHODS))
DBSCAN
OPTICS
DENCLUE

‫توز‬ ‫که‬ ‫هایی‬ ‫داده‬ ‫برای‬ ‫رود‬ ‫می‬ ‫کار‬ ‫به‬ ‫پیچیده‬ ‫و‬ ‫کروی‬ ‫غیر‬ ‫حاالت‬ ‫برای‬‫برخ‬ ‫قوی‬ ‫ریاضی‬ ‫بیس‬ ‫از‬ ‫البته‬ ‫و‬ ‫دارند‬ ‫ختی‬ ‫و‬ ‫کروی‬ ‫غیر‬ ‫یع‬‫وردارند‬

‫م‬ ‫ی‬ ‫لگ‬DBSCAN

N THIS DIAGRAM, MINPTS = 4. POINT A AND THE OTHER RED
POINTS ARE CORE POINTS, BECAUSE THE AREA
SURROUNDING THESE POINTS IN AN ΕRADIUS CONTAIN AT
LEAST 4 POINTS (INCLUDING THE POINT ITSELF). BECAUSE
THEY ARE ALL REACHABLE FROM ONE ANOTHER, THEY FORM
A SINGLE CLUSTER. POINTS B AND C ARE NOT CORE POINTS,
BUT ARE REACHABLE FROM A (VIA OTHER CORE POINTS)
AND THUS BELONG TO THE CLUSTER AS WELL. POINT N IS A
NOISE POINT THAT IS NEITHER A CORE POINT NOR DENSITY-
REACHABLE.‫کننده‬ ‫تنظیم‬:‫اسدی‬ ‫میثم‬ meysam_tabriz@yahoo.com

DBSCAN CAN FIND NON-LINEARLY SEPARABLE CLUSTERS.
THIS DATASET CANNOT BE ADEQUATELY CLUSTERED WITH
K-MEANS OR GAUSSIAN MIXTURE EM CLUSTERING.

OPTICS

‫شبیه‬ ‫خیلی‬DBSCAN‫مثل‬ ‫نداره‬ ‫پارامترها‬ ‫به‬ ‫حساسیت‬ ‫ولی‬MINPTS

DENCLUE

‫ی‬ ‫گ‬ ‫ج‬ ‫ط‬ ‫ک‬:(GRID-BASED METHODS)

STING

CLIQUE

:(PROBABILISTIC MODEL-BASED CLUSTERING)

SOM:Self-Organizing map
An illustration of the training of a self-organizing map. The blue blob is the
distribution of the training data, and the small white disc is the current
training datum drawn from that distribution. At first (left) the SOM nodes are
arbitrarily positioned in the data space. The node (highlighted in yellow)
which is nearest to the training datum is selected. It is moved towards the
training datum, as (to a lesser extent) are its neighbors on the grid. After
many iterations the grid tends to approximate the data distribution (right).

ONE-DIMENSIONAL SOM VERSUS PRINCIPAL COMPONENT ANALYSIS
(PCA) FOR DATA APPROXIMATION. SOM IS A REDBROKEN LINE WITH
SQUARES, 20 NODES. THE FIRST PRINCIPAL COMPONENT IS PRESENTED
BY A BLUE LINE. DATA POINTS ARE THE SMALL GREY CIRCLES. FOR PCA,
THEFRACTION OF VARIANCE UNEXPLAINED IN THIS EXAMPLE IS 23.23%,
FOR SOM IT IS 6.86%.[14]

Data mining

Recommended

Recommended

More Related Content

Viewers also liked

Viewers also liked (17)

Data mining