Data cleansing
● E.g. correct mistakes in addresses (misspellings, zip code errors)
● Merge address lists from different sources and purge duplicates
■ How to propagate updates
● Warehouse schema may be a (materialized) view of schema from
data sources
■ What data to summarize
● Raw data may be too large to store online
● Aggregate values (totals/subtotals) often suffice
● Queries on raw data can often be transformed by query optimizer
to use aggregate values
52 trang |
Chia sẻ: huyhoang44 | Lượt xem: 660 | Lượt tải: 0
Bạn đang xem trước 20 trang tài liệu Cơ sở dữ liệu - Chapter 18: Data analysis and mining, để xem tài liệu hoàn chỉnh bạn click vào nút DOWNLOAD ở trên
Database System Concepts
©Silberschatz, Korth and Sudarshan
See www.dbbook.com for conditions on reuse
Chapter 18: Data Analysis and Mining
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Chapter 18: Data Analysis and Mining
n Decision Support Systems
n Data Analysis and OLAP
n Data Warehousing
n Data Mining
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Decision Support Systems
n Decisionsupport systems are used to make business decisions, often
based on data collected by online transactionprocessing systems.
n Examples of business decisions:
l What items to stock?
l What insurance premium to change?
l To whom to send advertisements?
n Examples of data used for making decisions
l Retail sales transaction details
l Customer profiles (income, age, gender, etc.)
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
DecisionSupport Systems: Overview
n Data analysis tasks are simplified by specialized tools and SQL
extensions
l Example tasks
For each product category and each region, what were the total
sales in the last quarter and how do they compare with the same
quarter last year
As above, for each product category and each customer category
n Statistical analysis packages (e.g., : S++) can be interfaced with
databases
l Statistical analysis is a large field, but not covered here
n Data mining seeks to discover knowledge automatically in the form of
statistical rules and patterns from large databases.
n A data warehouse archives information gathered from multiple sources,
and stores it under a unified schema, at a single site.
l Important for large businesses that generate data from multiple
divisions, possibly at multiple sites
l Data may also be purchased externally
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Analysis and OLAP
n Online Analytical Processing (OLAP)
l Interactive analysis of data, allowing data to be summarized and
viewed in different ways in an online fashion (with negligible delay)
n Data that can be modeled as dimension attributes and measure
attributes are called multidimensional data.
l Measure attributes
measure some value
can be aggregated upon
e.g. the attribute number of the sales relation
l Dimension attributes
define the dimensions on which measure attributes (or
aggregates thereof) are viewed
e.g. the attributes item_name, color, and size of the sales
relation
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Cross Tabulation of sales by itemname
and color
n The table above is an example of a crosstabulation (crosstab), also
referred to as a pivottable.
l Values for one of the dimension attributes form the row headers
l Values for another dimension attribute form the column headers
l Other dimension attributes are listed on top
l Values in individual cells are (aggregates of) the values of the
dimension attributes that specify the cell.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Relational Representation of Crosstabs
n Crosstabs can be represented
as relations
n We use the value all is used to
represent aggregates
n The SQL:1999 standard
actually uses null values in
place of all despite confusion
with regular null values
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Cube
n A data cube is a multidimensional generalization of a crosstab
n Can have n dimensions; we show 3 below
n Crosstabs can be used as views on a data cube
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Online Analytical Processing
n Pivoting: changing the dimensions used in a crosstab is called
n Slicing: creating a crosstab for fixed values only
l Sometimes called dicing, particularly when values for multiple
dimensions are fixed.
n Rollup: moving from finergranularity data to a coarser granularity
n Drill down: The opposite operation that of moving from coarser
granularity data to finergranularity data
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Hierarchies on Dimensions
n Hierarchy on dimension attributes: lets dimensions to be viewed
at different levels of detail
H E.g. the dimension DateTime can be used to aggregate by hour of
day, date, day of week, month, quarter or year
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Cross Tabulation With Hierarchy
n Crosstabs can be easily extended to deal with hierarchies
H Can drill down or roll up on a hierarchy
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
OLAP Implementation
n The earliest OLAP systems used multidimensional arrays in memory to
store data cubes, and are referred to as multidimensional OLAP
(MOLAP) systems.
n OLAP implementations using only relational database features are called
relational OLAP (ROLAP) systems
n Hybrid systems, which store some summaries in memory and store the
base data and other summaries in a relational database, are called
hybrid OLAP (HOLAP) systems.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
OLAP Implementation (Cont.)
n Early OLAP systems precomputed all possible aggregates in order to
provide online response
l Space and time requirements for doing so can be very high
2n combinations of group by
l It suffices to precompute some aggregates, and compute others on
demand from one of the precomputed aggregates
Can compute aggregate on (itemname, color) from an aggregate
on (itemname, color, size)
– For all but a few “nondecomposable” aggregates such as
median
– is cheaper than computing it from scratch
n Several optimizations available for computing multiple aggregates
l Can compute aggregate on (itemname, color) from an aggregate on
(itemname, color, size)
l Can compute aggregates on (itemname, color, size),
(itemname, color) and (itemname) using a single sorting
of the base data
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Extended Aggregation in SQL:1999
n The cube operation computes union of group by’s on every subset of the
specified attributes
n E.g. consider the query
select itemname, color, size, sum(number)
from sales
group by cube(itemname, color, size)
This computes the union of eight different groupings of the sales relation:
{ (itemname, color, size), (itemname, color),
(itemname, size), (color, size),
(itemname), (color),
(size), ( ) }
where ( ) denotes an empty group by list.
n For each grouping, the result contains the null value
for attributes not present in the grouping.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Extended Aggregation (Cont.)
n Relational representation of crosstab that we saw earlier, but with null in
place of all, can be computed by
select itemname, color, sum(number)
from sales
group by cube(itemname, color)
n The function grouping() can be applied on an attribute
l Returns 1 if the value is a null value representing all, and returns 0 in all
other cases.
select itemname, color, size, sum(number),
grouping(itemname) as itemnameflag,
grouping(color) as colorflag,
grouping(size) as sizeflag,
from sales
group by cube(itemname, color, size)
n Can use the function decode() in the select clause to replace
such nulls by a value such as all
l E.g. replace itemname in first query by
decode( grouping(itemname), 1, ‘all’, itemname)
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Extended Aggregation (Cont.)
n The rollup construct generates union on every prefix of specified list of
attributes
n E.g.
select itemname, color, size, sum(number)
from sales
group by rollup(itemname, color, size)
Generates union of four groupings:
{ (itemname, color, size), (itemname, color), (itemname), ( ) }
n Rollup can be used to generate aggregates at multiple levels of a
hierarchy.
n E.g., suppose table itemcategory(itemname, category) gives the
category of each item. Then
select category, itemname, sum(number)
from sales, itemcategory
where sales.itemname = itemcategory.itemname
group by rollup(category, itemname)
would give a hierarchical summary by itemname and by category.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Extended Aggregation (Cont.)
n Multiple rollups and cubes can be used in a single group by clause
l Each generates set of group by lists, cross product of sets gives overall
set of group by lists
n E.g.,
select itemname, color, size, sum(number)
from sales
group by rollup(itemname), rollup(color, size)
generates the groupings
{itemname, ()} X {(color, size), (color), ()}
= { (itemname, color, size), (itemname, color), (itemname),
(color, size), (color), ( ) }
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Ranking
n Ranking is done in conjunction with an order by specification.
n Given a relation studentmarks(studentid, marks) find the rank of each
student.
select studentid, rank( ) over (order by marks desc) as srank
from studentmarks
n An extra order by clause is needed to get them in sorted order
select studentid, rank ( ) over (order by marks desc) as srank
from studentmarks
order by srank
n Ranking may leave gaps: e.g. if 2 students have the same top mark, both
have rank 1, and the next rank is 3
l dense_rank does not leave gaps, so next dense rank would be 2
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Ranking (Cont.)
n Ranking can be done within partition of the data.
n “Find the rank of students within each section.”
select studentid, section,
rank ( ) over (partition by section order by marks desc)
as secrank
from studentmarks, studentsection
where studentmarks.studentid = studentsection.studentid
order by section, secrank
n Multiple rank clauses can occur in a single select clause
n Ranking is done after applying group by clause/aggregation
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Ranking (Cont.)
n Other ranking functions:
l percent_rank (within partition, if partitioning is done)
l cume_dist (cumulative distribution)
fraction of tuples with preceding values
l row_number (nondeterministic in presence of duplicates)
n SQL:1999 permits the user to specify nulls first or nulls last
select studentid,
rank ( ) over (order by marks desc nulls last) as srank
from studentmarks
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Ranking (Cont.)
n For a given constant n, the ranking the function ntile(n) takes the
tuples in each partition in the specified order, and divides them into n
buckets with equal numbers of tuples.
n E.g.:
select threetile, sum(salary)
from (
select salary, ntile(3) over (order by salary) as threetile
from employee) as s
group by threetile
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Windowing
n Used to smooth out random variations.
n E.g.: moving average: “Given sales values for each date, calculate for each
date the average of the sales on that day, the previous day, and the next
day”
n Window specification in SQL:
l Given relation sales(date, value)
select date, sum(value) over
(order by date between rows 1 preceding and 1 following)
from sales
n Examples of other window specifications:
l between rows unbounded preceding and current
l rows unbounded preceding
l range between 10 preceding and current row
All rows with values between current row value –10 to current value
l range interval 10 day preceding
Not including current row
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Windowing (Cont.)
n Can do windowing within partitions
n E.g. Given a relation transaction (accountnumber, datetime, value),
where value is positive for a deposit and negative for a withdrawal
l “Find total balance of each account after each transaction on the
account”
select accountnumber, datetime,
sum (value ) over
(partition by accountnumber
order by datetime
rows unbounded preceding)
as balance
from transaction
order by accountnumber, datetime
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Warehousing
n Data sources often store only current data, not historical data
n Corporate decision making requires a unified view of all organizational
data, including historical data
n A data warehouse is a repository (archive) of information gathered
from multiple sources, stored under a unified schema, at a single site
l Greatly simplifies querying, permits study of historical trends
l Shifts decision support query load away from transaction
processing systems
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Warehousing
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Design Issues
n When and how to gather data
l Source driven architecture: data sources transmit new information
to warehouse, either continuously or periodically (e.g. at night)
l Destination driven architecture: warehouse periodically requests
new information from data sources
l Keeping warehouse exactly synchronized with data sources (e.g.
using twophase commit) is too expensive
Usually OK to have slightly outofdate data at warehouse
Data/updates are periodically downloaded form online
transaction processing (OLTP) systems.
n What schema to use
l Schema integration
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
More Warehouse Design Issues
n Data cleansing
l E.g. correct mistakes in addresses (misspellings, zip code errors)
l Merge address lists from different sources and purge duplicates
n How to propagate updates
l Warehouse schema may be a (materialized) view of schema from
data sources
n What data to summarize
l Raw data may be too large to store online
l Aggregate values (totals/subtotals) often suffice
l Queries on raw data can often be transformed by query optimizer
to use aggregate values
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Warehouse Schemas
n Dimension values are usually encoded using small integers and
mapped to full values via dimension tables
n Resultant schema is called a star schema
l More complicated schema structures
Snowflake schema: multiple levels of dimension tables
Constellation: multiple fact tables
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Warehouse Schema
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Mining
n Data mining is the process of semiautomatically analyzing large
databases to find useful patterns
n Prediction based on past history
l Predict if a credit card applicant poses a good credit risk, based on
some attributes (income, job type, age, ..) and past history
l Predict if a pattern of phone calling card usage is likely to be
fraudulent
n Some examples of prediction mechanisms:
l Classification
Given a new item whose class is unknown, predict to which class
it belongs
l Regression formulae
Given a set of mappings for an unknown function, predict the
function result for a new parameter value
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Data Mining (Cont.)
n Descriptive Patterns
l Associations
Find books that are often bought by “similar” customers. If a
new such customer buys one such book, suggest the others
too.
l Associations may be used as a first step in detecting causation
E.g. association between exposure to chemical X and cancer,
l Clusters
E.g. typhoid cases were clustered in an area surrounding a
contaminated well
Detection of clusters remains important in detecting epidemics
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Classification Rules
n Classification rules help assign new objects to classes.
l E.g., given a new automobile insurance applicant, should he or she
be classified as low risk, medium risk or high risk?
n Classification rules for above example could use a variety of data, such
as educational level, salary, age, etc.
l ∀ person P, P.degree = masters and P.income > 75,000
⇒ P.credit = excellent
l ∀ person P, P.degree = bachelors and
(P.income ≥ 25,000 and P.income ≤ 75,000)
⇒ P.credit = good
n Rules are not necessarily exact: there may be some misclassifications
n Classification rules can be shown compactly as a decision tree.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Decision Tree
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Construction of Decision Trees
n Training set: a data sample in which the classification is already
known.
n Greedy top down generation of decision trees.
l Each internal node of the tree partitions the data into groups
based on a partitioning attribute, and a partitioning condition
for the node
l Leaf node:
all (or most) of the items at the node belong to the same class,
or
all attributes have been considered, and no further partitioning
is possible.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Best Splits
n Pick best attributes and conditions on which to partition
n The purity of a set S of training instances can be measured quantitatively in
several ways.
l Notation: number of classes = k, number of instances = |S|,
fraction of instances in class i = pi.
n The Gini measure of purity is defined as
[
Gini (S) = 1 ∑
l When all instances are in a single class, the Gini value is 0
l It reaches its maximum (of 1 –1 /k) if each class the same number of
instances.
k
i 1
p2i
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Best Splits (Cont.)
n Another measure of purity is the entropy measure, which is defined as
entropy (S) = – ∑
n When a set S is split into multiple sets Si, I=1, 2, , r, we can measure the
purity of the resultant set of sets as:
purity(S1, S2, .., Sr) = ∑
n The information gain due to particular split of S into Si, i = 1, 2, ., r
Informationgain (S, {S1, S2, ., Sr) = purity(S ) – purity (S1, S2, Sr)
r
i= 1
|Si|
|S|
purity (Si)
k
i 1
pilog2 pi
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Best Splits (Cont.)
n Measure of “cost” of a split:
Informationcontent (S, {S1, S2, .., Sr})) = – ∑
n Informationgain ratio = Informationgain (S, {S1, S2, , Sr})
Informationcontent (S, {S1, S2, .., Sr})
n The best split is the one that gives the maximum information gain ratio
log2
r
i 1
|Si|
|S|
|Si|
|S|
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Finding Best Splits
n Categorical attributes (with no meaningful order):
l Multiway split, one child for each value
l Binary split: try all possible breakup of values into two sets, and
pick the best
n Continuousvalued attributes (can be sorted in a meaningful order)
l Binary split:
Sort values, try each as a split point
– E.g. if values are 1, 10, 15, 25, split at ≤1, ≤ 10, ≤ 15
Pick the value that gives best split
l Multiway split:
A series of binary splits on the same attribute has roughly
equivalent effect
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
DecisionTree Construction Algorithm
Procedure GrowTree (S )
Partition (S );
Procedure Partition (S)
if ( purity (S ) > δp or |S| < δs ) then
return;
for each attribute A
evaluate splits on attribute A;
Use best split found (across all attributes) to partition
S into S1, S2, ., Sr,
for i = 1, 2, .., r
Partition (Si );
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Other Types of Classifiers
n Neural net classifiers are studied in artificial intelligence and are not covered
here
n Bayesian classifiers use Bayes theorem, which says
p (cj | d ) = p (d | cj ) p (cj )
p ( d )
where
p (cj | d ) = probability of instance d being in class cj,
p (d | cj ) = probability of generating instance d given class cj,
p (cj ) = probability of occurrence of class cj, and
p (d ) = probability of instance d occuring
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Naïve Bayesian Classifiers
n Bayesian classifiers require
l computation of p (d | cj )
l precomputation of p (cj )
l p (d ) can be ignored since it is the same for all classes
n To simplify the task, naïve Bayesian classifiers assume attributes
have independent distributions, and thereby estimate
p (d | cj) = p (d1 | cj ) * p (d2 | cj ) * .* (p (dn | cj )
l Each of the p (di | cj ) can be estimated from a histogram on di
values for each class cj
the histogram is computed from the training instances
l Histograms on multiple attributes are more expensive to compute
and store
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Regression
n Regression deals with the prediction of a value, rather than a class.
l Given values for a set of variables, X1, X2, , Xn, we wish to predict the
value of a variable Y.
n One way is to infer coefficients a0, a1, a1, , an such that
Y = a0 + a1 * X1 + a2 * X2 + + an * Xn
n Finding such a linear polynomial is called linear regression.
l In general, the process of finding a curve that fits the data is also called
curve fitting.
n The fit may only be approximate
l because of noise in the data, or
l because the relationship is not exactly a polynomial
n Regression aims to find coefficients that give the best possible fit.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Association Rules
n Retail shops are often interested in associations between different items
that people buy.
l Someone who buys bread is quite likely also to buy milk
l A person who bought the book Database System Concepts is quite
likely also to buy the book Operating System Concepts.
n Associations information can be used in several ways.
l E.g. when a customer buys a particular book, an online shop may
suggest associated books.
n Association rules:
bread ⇒ milk DBConcepts, OSConcepts ⇒ Networks
l Left hand side: antecedent, right hand side: consequent
l An association rule must have an associated population; the
population consists of a set of instances
E.g. each transaction (sale) at a shop is an instance, and the set
of all transactions is the population
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Association Rules (Cont.)
n Rules have an associated support, as well as an associated confidence.
n Support is a measure of what fraction of the population satisfies both the
antecedent and the consequent of the rule.
l E.g. suppose only 0.001 percent of all purchases include milk and
screwdrivers. The support for the rule is milk ⇒ screwdrivers is low.
n Confidence is a measure of how often the consequent is true when the
antecedent is true.
l E.g. the rule bread ⇒ milk has a confidence of 80 percent if 80
percent of the purchases that include bread also include milk.
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Finding Association Rules
n We are generally only interested in association rules with reasonably
high support (e.g. support of 2% or greater)
n Naïve algorithm
1. Consider all possible sets of relevant items.
2. For each set find its support (i.e. count how many transactions
purchase all items in the set).
H Large itemsets: sets with sufficiently high support
3. Use large itemsets to generate association rules.
1. From itemset A generate the rule A {b } ⇒b for each b ∈ A.
4 Support of rule = support (A).
4 Confidence of rule = support (A ) / support (A {b })
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Finding Support
n Determine support of itemsets via a single pass on set of transactions
l Large itemsets: sets with a high count at the end of the pass
n If memory not enough to hold all counts for all itemsets use multiple passes,
considering only some itemsets in each pass.
n Optimization: Once an itemset is eliminated because its count (support) is too
small none of its supersets needs to be considered.
n The a priori technique to find large itemsets:
l Pass 1: count support of all sets with just 1 item. Eliminate those items
with low support
l Pass i: candidates: every set of i items such that all its i1 item subsets
are large
Count support of all candidates
Stop if there are no candidates
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Other Types of Associations
n Basic association rules have several limitations
n Deviations from the expected probability are more interesting
l E.g. if many people purchase bread, and many people purchase cereal,
quite a few would be expected to purchase both
l We are interested in positive as well as negative correlations between
sets of items
Positive correlation: cooccurrence is higher than predicted
Negative correlation: cooccurrence is lower than predicted
n Sequence associations / correlations
l E.g. whenever bonds go up, stock prices go down in 2 days
n Deviations from temporal patterns
l E.g. deviation from a steady growth
l E.g. sales of winter wear go down in summer
Not surprising, part of a known pattern.
Look for deviation from value predicted using past patterns
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Clustering
n Clustering: Intuitively, finding clusters of points in the given data such that
similar points lie in the same cluster
n Can be formalized using distance metrics in several ways
l Group points into k sets (for a given k) such that the average distance
of points from the centroid of their assigned group is minimized
Centroid: point defined by taking average of coordinates in each
dimension.
l Another metric: minimize average distance between every pair of
points in a cluster
n Has been studied extensively in statistics, but on small data sets
l Data mining systems aim at clustering techniques that can handle very
large data sets
l E.g. the Birch clustering algorithm (more shortly)
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Hierarchical Clustering
n Example from biological classification
l (the word classification here does not mean a prediction mechanism)
chordata
mammalia reptilia
leopards humans snakes crocodiles
n Other examples: Internet directory systems (e.g. Yahoo, more on this later)
n Agglomerative clustering algorithms
l Build small clusters, then cluster small clusters into bigger clusters, and
so on
n Divisive clustering algorithms
l Start with all items in a single cluster, repeatedly refine (break) clusters
into smaller ones
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Clustering Algorithms
n Clustering algorithms have been designed to handle very large
datasets
n E.g. the Birch algorithm
l Main idea: use an inmemory Rtree to store points that are being
clustered
l Insert points one at a time into the Rtree, merging a new point
with an existing cluster if is less than some δ distance away
l If there are more leaf nodes than fit in memory, merge existing
clusters that are close to each other
l At the end of first pass we get a large number of clusters at the
leaves of the Rtree
Merge clusters to reduce the number of clusters
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Collaborative Filtering
n Goal: predict what movies/books/ a person may be interested in, on
the basis of
l Past preferences of the person
l Other people with similar past preferences
l The preferences of such people for a new movie/book/
n One approach based on repeated clustering
l Cluster people on the basis of preferences for movies
l Then cluster movies on the basis of being liked by the same
clusters of people
l Again cluster people based on their preferences for (the newly
created clusters of) movies
l Repeat above till equilibrium
n Above problem is an instance of collaborative filtering, where users
collaborate in the task of filtering information to find information of
interest
©Silberschatz, Korth and Sudarshan18.Database System Concepts 5th Edition, Aug 26, 2005
Other Types of Mining
n Text mining: application of data mining to textual documents
l cluster Web pages to find related pages
l cluster pages a user has visited to organize their visit history
l classify Web pages automatically into a Web directory
n Data visualization systems help users examine large volumes of data
and detect patterns visually
l Can visually encode large amounts of information on a single
screen
l Humans are very good a detecting visual patterns
Các file đính kèm theo tài liệu này:
- ch18_3656_6518.pdf