[go: up one dir, main page]
Scribd
Upload
11 views13 pages

106 Unsupervised Learning - Association Rules

Uploaded by

omshewale18032003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
11 views13 pages

106 Unsupervised Learning - Association Rules

Uploaded by

omshewale18032003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 13

106

UNSUPERVISE
D LEARNING –
A SS O C I AT I O N
RULES

S U N I L G O RA N T L A
INTRODUCTION
• A method for discovering interesting relationships (associations)
between variables in large datasets.
• Commonly used in market basket analysis to identify products
that frequently co-occur in transactions.
• Example: If a customer buys bread, they are likely to buy butter.
FREQU ENCY AND SU P P ORT
• Frequent Itemsets:
• An itemset is a collection of one or more items.
• An itemset is considered frequent if it appears in the dataset
at least as many times as a pre-specified threshold (minimum
support).
• Support:
• The proportion of transactions that contain the itemset.
• Formula:
CONFIDENCE, LIFT
• Confidence:
• The likelihood that a transaction containing itemset X also contains
itemset Y.

• Lift:
• The ratio of the observed support to the expected support under
independence.

• Lift > 1 indicates a positive association between X and Y.


• Total = 100

• Tomato = 100
• Bread = 40
• Jam = 15
ALGORITHM
• Step 1: Generate candidate itemsets of length 1.
• Step 2: Prune the itemsets that do not meet the minimum
support threshold.
• Step 3: Generate candidate itemsets of length 2 from the
frequent itemsets of length 1.
• Step 4: Repeat until no more frequent itemsets can be
generated.
• Step 5: Generate association rules from the frequent itemsets.
E C L AT
• ECLAT (Equivalence Class Clustering and bottom-up Lattice
Traversal) is an efficient algorithm for finding frequent itemsets.
• Unlike Apriori, which uses horizontal data layout (itemsets),
ECLAT uses vertical data layout (transaction IDs).
• Developed as an alternative to Apriori to handle large datasets
more efficiently.
T I D AND SU P P ORT
• Transaction ID Sets (TID Sets):In ECLAT, each item is
represented by the set of transaction IDs where it appears.
• The algorithm computes intersections of TID sets to find frequent
itemsets.
• Support:
• The support of an itemset is determined by the size of its TID
set.
ALGORITHM
• Step 1: Convert the dataset into a vertical format where each item is
associated with a TID set.
• Step 2: Calculate the support for individual items by counting the
number of transactions in their TID sets.
• Step 3: Generate larger itemsets by intersecting TID sets of smaller
itemsets (e.g., pairwise intersections for 2-itemsets).
• Step 4: Prune itemsets that do not meet the minimum support threshold.
• Step 5: Continue this process until no larger frequent itemsets can be
generated.
Items
Transaction ID Purchased
{Bread, Milk,
T1 Butter}
T2 {Bread, Milk}
EXAMPLE
T3 {Milk, Butter}
T4 {Bread, Butter}
{Bread, Milk,
T5 Butter}
D A T A L AY O U T
• Apriori • ECLAT
• T1: {Bread, Milk, Butter} • Bread: {T1, T2, T4, T5}
• T2: {Bread, Milk} • Milk: {T1, T2, T3, T5}
• T3: {Milk, Butter} • Butter: {T1, T3, T4, T5}
• T4: {Bread, Butter}
• T5: {Bread, Milk, Butter}
FREQUENT 1-ITEMSETS
G E N E R AT I O N
• Apriori • ECLAT
• {Bread}: 4 transactions • Bread: {T1, T2, T4, T5} (4
• {Milk}: 4 transactions transactions)
• {Butter}: 4 transactions • Milk: {T1, T2, T3, T5} (4
transactions)
• Butter: {T1, T3, T4, T5} (4
transactions)
FREQUENT 2-ITEMSETS
G E N E R AT I O N
• Apriori • ECLAT
• {Bread, Milk}: 3 • {Bread, Milk}: {T1, T2, T5} (3
transactions transactions)
• {Bread, Butter}: 3 • {Bread, Butter}: {T1, T4, T5} (3
transactions transactions)
• {Milk, Butter}: 3 • {Milk, Butter}: {T1, T3, T5} (3
transactions transactions)

You might also like