Frequent Pattern &
Sequential Pattern Mining
NAME:SRIJITA NAG
ROLL NO : 38732523024
DEPT:ARTIFICIAL INTELLIGENCE
SUBJECT :DATA MINING
SUBJECT CODE : PEC-AI 601B
Introduction to Data Mining
Data mining is the process of extracting useful information and
meaningful patterns from large volumes of data. With the rapid
growth of digital data, it has become difficult to analyze data
manually. Data mining helps in discovering hidden patterns,
relationships, and trends that are useful for decision making.
Key points of data mining are:
Works on large and complex datasets
Converts raw data into useful knowledge
Helps in prediction and analysis
Used in business, healthcare, education, and banking
Includes tasks avoiding like pattern mining, classification, and
clustering
Frequent pattern mining is a data mining technique
Frequent Pattern Mining used to discover patterns that occur frequently in a
dataset. These patterns show relationships among
items and help in understanding data behavior. It is
commonly used in market basket analysis to study
customer buying habits.
Key points of frequent pattern mining are:
Identifies frequently occurring itemsets
Helps discover hidden relationships in data
Forms the basis of association rule mining
Useful in business and recommendation systems
Improves decision-making by analyzing patterns
Frequent pattern mining helps organizations analyze
data efficiently and make better predictions.
Types of Frequent Patterns
Frequent patterns can be classified based on the structure of
data and the type of relationships present among items.
Identifying different types of frequent patterns helps in reducing
redundancy and improving the efficiency of the mining process.
Main types of frequent patterns are:
Frequent Itemsets – Sets of items that appear together
frequently
Frequent Subsequences – Ordered sequences that occur
often
Frequent Substructures – Common structures like graphs or
trees
Closed Frequent Patterns – Patterns with no super-pattern
having same support
Maximal Frequent Patterns – Largest frequent patterns in a
dataset
These pattern types help in effective and optimized data
analysis.
Association Rule Mining
Association rule mining is a technique used to discover
relationships between items in large datasets. It helps in
identifying how items are connected or occur together.
This method is widely used in market basket analysis to
understand customer purchasing behavior.
Key points of association rule mining are:
Discovers relationships among items
Expressed in the form X→ Y
Based on frequent itemsets
Helps in product recommendation
Used in retail and e-commerce systems
Association rule mining helps organizations make better
marketing and business decisions.
Measures of Association Rules
SUPPORT CONFIDENCE
Shows how frequently an itemset Indicates how often the rule is
appears in the dataset true
LIFT
Measures the strength of
association between items
Correlation
Correlation analysis is used to determine whether
items are truly related to each other or just appear
Analysis
together by chance. It helps in identifying the real
strength of association between items in a dataset.
This analysis avoids misleading rules generated
only on the basis of support and confidence.
Key points of correlation analysis are:
Identifies positive and negative relationships
Measures true dependency between items
Helps remove irrelevant association rules
Improves accuracy of pattern mining
Uses measures like lift and correlation
coefficient
Sequential Pattern Mining
Sequential pattern mining focuses on
discovering frequent sequences in data where
the order of events matters. Unlike frequent
itemset mining, it considers the time or sequence
in which data occurs. This technique is useful for
analyzing behavior that changes over time.
Why sequential pattern mining is important:
Captures time-based data patterns
Identifies frequent event sequences
Helps analyze customer behavior over time
Used in web navigation and purchase history
Supports prediction of future actions
Pattern Mining Primitives
Pattern mining primitives define what kind of
patterns are to be discovered and how the
mining process should be performed. They act
as guidelines that control the mining task and
make the process more efficient and meaningful
Key pattern mining primitives include:
Type of patterns to be mined
Minimum support threshold
Measures of interestingness
Constraints on pattern selection
Form of output patterns
.
TYPES OF PATTERNS TO BE MINED
FREQUENT ITEMSETS SEQUENTIAL PATTERNS
Items that occur together Ordered sequences of
frequently events or items
ASSOCIATION RULES CORRELATED PATTERNS
Relationships between items Patterns showing true dependency
in the form X→ Y
CONSTRAINTS IN PATTERN MINING
Constraints in pattern mining are used to limit the search space and focus only on useful patterns. By applying
constraints, the mining process becomes faster and more efficient, especially when working with large datasets.
ANTI-MONOTONIC
MONOTONIC CONSTRAINTS SUCCINCT CONSTRAINTS
CONSTRAINTS
If a pattern satisfies the constraint, Applied directly during data
If a pattern is infrequent, its
its supersets also satisfy it scanning
supersets are also infrequent
CONVERTIBLE CONSTRAINTS
Can act as monotonic or anti-
monotonic after ordering
Scalable Methods in Data Mining
Scalable methods are designed to handle very
large datasets efficiently. As data size grows,
traditional mining techniques become slow and
inefficient. Scalable methods ensure that pattern
mining remains practical and fast.
Key features of scalable methods include:
Efficient use of memory
Reduced computation time
Ability to process large databases
Use of parallel and distributed computing
Support for incremental data mining
TYPES OF SCALABLE METHODS
DATABASE-ORIENTED METHODS SAMPLING-BASED METHODS
Reduce database scans and Analyze representative
memory usage samples instead of full data
PARALLEL AND DISTRIBUTED METHODS INCREMENTAL MINING METHODS
Use multiple processors or Update patterns when new
systems data arrives
Pattern mining techniques are widely used to discover Applications of Pattern
useful knowledge from large datasets. These applications
help organizations understand behavior, improve Mining
services, and make better decisions.
Major applications include:
Market basket analysis – Identifies customer buying
patterns
Recommendation systems – Suggests products or
content
Fraud detection – Detects unusual or suspicious
activities
Web usage mining – Analyzes user navigation
behavior
Healthcare analysis – Studies medical and treatment
patterns
Pattern mining plays a crucial role in data-driven
decision making across various domains.
Conclusion
Frequent pattern mining and sequential pattern mining help in discovering
meaningful relationships from large datasets. Association and correlation analysis
improve the quality of patterns, while primitives and scalable methods make the
mining process efficient. These techniques are widely used in real-life applications
such as business analysis, web mining, and healthcare, supporting effective data-
driven decision making.
Thank You