DATA MINING

LECTURE 2
Frequent Itemsets, Association Rules
 Outline
• Market-Basket Data
• Frequent Itemsets
• Applications
• Mining Frequent Itemsets
• Itemset lattice
• A Naïve Algorithm
• The Apriori Principle
• The Apriori algorithm
• Examples
• Hash tree
• Association Rule Mining
• http://www.philippe-fournier-viger.com/spmf/Apriori.php
• http://www.philippe-fournier-viger.com/spmf/AprioriTID.php
 3

Market-Basket Data
• A large set of items, e.g., things sold in a
supermarket.
• A large set of baskets, each of which is a small
set of the items, e.g., the things one customer
buys on one day.
 4

Market-Baskets – (2)
• Really, a general many-to-many mapping
(association) between two kinds of things, where
the one (the baskets) is a set of the other (the
items)
• But we ask about connections among “items,” not
“baskets.”
• The technology focuses on common events, not
rare events (“long tail”).
 Frequent Itemsets
• Given a set of transactions, find combinations of items
(itemsets) that occur frequently
Support 𝑠 𝐼 : number of
transactions that contain
Market-Basket transactions
itemset I
Items: {Bread, Milk, Diaper, Beer, Eggs, Coke}
TID Items Examples of frequent itemsets 𝑠 𝐼 ≥ 3
1 Bread, Milk
2 Bread, Diaper, Beer, Eggs {Bread}: 4
{Milk} : 4
3 Milk, Diaper, Beer, Coke {Diaper} : 4
4 Bread, Milk, Diaper, Beer {Beer}: 3
5 Bread, Milk, Diaper, Coke {Diaper, Beer} : 3
{Milk, Bread} : 3
 6

Applications – (1)
• Items = products; baskets = sets of products
someone bought in one trip to the store.

• Example application: given that many people buy

beer and diapers together:
• Run a sale on diapers; raise price of beer.
• Only useful if many buy diapers & beer.
 7

Applications – (2)
• Baskets = Web pages; items = words.

• Example application: words appearing together

in a large number of documents, e.g., “Brad” and
“Angelina,” may indicate an interesting
relationship.
 8

Applications – (3)
• Baskets = sentences; items = documents
containing those sentences.

• Example application: Items that appear together

too often could represent plagiarism.
 Definition: Frequent Itemset
• Itemset
• A collection of one or more items
• Example: {Milk, Bread, Diaper}
• k-itemset TID Items
• An itemset that contains k items 1 Bread, Milk
• Support () 2 Bread, Diaper, Beer, Eggs
• Count: Frequency of occurrence of an 3 Milk, Diaper, Beer, Coke
itemset 4 Bread, Milk, Diaper, Beer
• E.g. ({Milk, Bread,Diaper}) = 2 5 Bread, Milk, Diaper, Coke
• Fraction: Fraction of transactions that
contain an itemset
• E.g. s({Milk, Bread, Diaper}) = 40%
• Frequent Itemset
• An itemset whose support is greater 𝑠 𝐼 ≥ minsup
than or equal to a minsup threshold
Mining Frequent Itemsets task
• Input: A set of transactions T, over a set of items I
• Output: All itemsets with items in I having
• support ≥ minsup threshold

• Problem parameters:
• N = |T|: number of transactions
• d = |I|: number of (distinct) items
• w: max width of a transaction
• Number of possible itemsets? M = 2d

• Scale of the problem:

• WalMart sells 100,000 items and can store billions of baskets.
• The Web has billions of words and many billions of pages.
 The itemset lattice
null

A B C D E

AB AC AD AE BC BD BE CD CE DE

ABC ABD ABE ACD ACE ADE BCD BCE BDE CDE

ABCD ABCE ABDE ACDE BCDE

Given d items, there are

ABCDE 2d possible itemsets
 A Naïve Algorithm
• Brute-force approach, each itemset is a candidate :
• Consider each itemset in the lattice, and count the support of each candidate by
scanning the data
• Time Complexity ~ O(NMw) , Space Complexity ~ O(M)
• OR
• Scan the data, and for each transaction generate all possible itemsets. Keep a count
for each itemset in the data.
• Time Complexity ~ O(N2ww) , Space Complexity ~ O(M)

• Expensive since M = 2d !!!

Transactions List of
Candidates
TID Items
1 Bread, Milk
2 Bread, Diaper, Beer, Eggs
N 3 Milk, Diaper, Beer, Coke M
4 Bread, Milk, Diaper, Beer
5 Bread, Milk, Diaper, Coke
w
Example file: retail
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
30 31 32
33 34 35
36 37 38 39 40 41 42 43 44 45 46
38 39 47 48
38 39 48 49 50 51 52 53 54 55 56 57 58 Example:
32 41 59 60 61 62
3 39 48
- Items are positive integers,
63 64 65 66 67 68 - and each basket corresponds to a line
32 69
in the file of space separated integers
48 70 71 72
39 73 74 75 76 77 78 79
36 38 39 41 48 79 80 81
82 83 84
41 85 86 87 88
39 48 89 90 91 92 93 94 95 96 97 98 99 100 101
36 38 39 48 89
39 41 102 103 104 105 106 107 108
38 39 41 109 110
39 111 112 113 114 115 116 117 118
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
48 134 135 136
39 48 137 138 139 140 141 142 143 144 145 146 147 148 149
39 150 151 152
38 39 56 153 154 155
The Apriori Principle
• Apriori principle (Main observation):
– If an itemset is frequent, then all of its subsets must also
be frequent
– If an itemset is not frequent, then all of its supersets
cannot be frequent
X , Y : ( X  Y )  s ( X )  s (Y )
– The support of an itemset never exceeds the support of
its subsets
– This is known as the anti-monotone property of support
Illustration of the Apriori principle
Frequent
subsets

Found to be frequent
 Illustration of the Apriori principle
null

A B C D E

AB AC AD AE BC BD BE CD CE DE

Found to be
Infrequent
ABC ABD ABE ACD ACE ADE BCD BCE BDE CDE

ABCD ABCE ABDE ACDE BCDE

Infrequent supersets
ABCDE
Pruned
 The Apriori algorithm
Ck = candidate itemsets of size k
Level-wise approach Lk = frequent itemsets of size k

1. k = 1, C1 = all items
2. While Ck not empty
Frequent 3. Scan the database to find which itemsets in
itemset
generation Ck are frequent and put them into Lk
Candidate 4. Use Lk to generate a collection of candidate
generation itemsets Ck+1 of size k+1
5. k = k+1
R. Agrawal, R. Srikant: "Fast Algorithms for Mining Association Rules",
Proc. of the 20th Int'l Conference on Very Large Databases, 1994.
Illustration of the Apriori principle
TID Items
1 Bread, Milk

minsup = 3 2
3
Bread, Diaper, Beer, Eggs
Milk, Diaper, Beer, Coke
Item Count Items (1-itemsets) 4 Bread, Milk, Diaper, Beer
Bread 4 5 Bread, Milk, Diaper, Coke
Coke 2
Milk 4 Itemset Count Pairs (2-itemsets)
Beer 3
{Bread,Milk} 3
Diaper 4
{Bread,Beer} 2 (No need to generate
Eggs 1
{Bread,Diaper} 3 candidates involving Coke
{Milk,Beer} 2 or Eggs)
{Milk,Diaper} 3
{Beer,Diaper} 3
Triplets (3-itemsets)
If every subset is considered,
6 6 6 Itemset Count
+ + = 6 + 15 + 20 = 41 {Bread,Milk,Diaper} 2
1 2 3
With support-based pruning,
6 4 Only this triplet has all subsets to be frequent
+ + 1 = 6 + 6 + 1 = 13
1 2 But it is below the minsup threshold
Candidate Generation
• Basic principle (Apriori):
• An itemset of size k+1 is candidate to be frequent only if
all of its subsets of size k are known to be frequent
• Main idea:
• Construct a candidate of size k+1 by combining
frequent itemsets of size k
• If k = 1, take the all pairs of frequent items
• If k > 1, join pairs of itemsets that differ by just one item
• For each generated candidate itemset ensure that all subsets of
size k are frequent.
Generate Candidates Ck+1
• Assumption: The items in an itemset are ordered
• E.g., if integers ordered in increasing order, if strings ordered in
lexicographic order
• The items in Lk are also listed in an order

Create a candidate itemset of size k+1, by joining

two itemsets of size k, that share the first k-1 items
Item 1 Item 2 Item 3
1 2 3
1 2 5
1 4 5
Generate Candidates Ck+1
• Assumption: The items in an itemset are ordered
• E.g., if integers ordered in increasing order, if strings ordered in
lexicographic order
• The order ensures that if item y > x appears before x, then x is not in the
itemset

• The items in Lk are also listed in an order

Create a candidate itemset of size k+1, by joining

two itemsets of size k, that share the first k-1 items
Item 1 Item 2 Item 3
1 2 3
1 2 3 5
1 2 5
1 4 5
Generate Candidates Ck+1
• Assumption: The items in an itemset are ordered
• E.g., if integers ordered in increasing order, if strings ordered in
lexicographic order
• The order ensures that if item y > x appears before x, then x is not in the
itemset

• The items in Lk are also listed in an order

Create a candidate itemset of size k+1, by joining

two itemsets of size k, that share the first k-1 items
Item 1 Item 2 Item 3
Are we missing something?
1 2 3
What about this candidate?
1 2 5
1 2 4 5
1 4 5
Generating Candidates Ck+1 in SQL

• self-join Lk
insert into Ck+1
select p.item1, p.item2, …, p.itemk, q.itemk
from Lk p, Lk q
where p.item1=q.item1, …, p.itemk-1=q.itemk-1, p.itemk < q.itemk
Example I
• L3={abc, abd, acd, ace, bcd}
• Self-join: L3*L3
– abcd from abc and abd
– acde from acd and ace

item1 item2 item3 item1 item2 item3

a b c a b c
a b d a b d
a c d a c d
a c e a c e
b c d b c d

p.item1=q.item1,p.item2=q.item2, p.item3< q.item3

Example I
• L3={abc, abd, acd, ace, bcd}
• Self-joining: L3*L3
– abcd from abc and abd
– acde from acd and ace

item1 item2 item3 item1 item2 item3

a b c a b c
a b d a b d
a c d a c d
a c e a c e
b c d b c d

p.item1=q.item1,p.item2=q.item2, p.item3< q.item3

 Example I
• L3={abc, abd, acd, ace, bcd}
• Self-joining: L3*L3
– abcd from abc and abd
– acde from acd and ace
item1 item2 item3 item1 item2 item3
a b c a b c
a b d a b d
{a,b,c} {a,b,d}
a c d a c d
a c e a c e
{a,b,c,d}
b c d b c d

p.item1=q.item1,p.item2=q.item2, p.item3< q.item3

 Example I
• L3={abc, abd, acd, ace, bcd}
• Self-joining: L3*L3
– abcd from abc and abd
– acde from acd and ace
item1 item2 item3 item1 item2 item3
a b c a b c
a b d a b d {a,c,d} {a,c,e}
a c d a c d
a c e a c e {a,c,d,e}
b c d b c d

p.item1=q.item1,p.item2=q.item2, p.item3< q.item3

Example II
Itemset Count
{Beer,Diaper} 3
{Bread,Diaper} 3
{Bread,Milk} 3
{Diaper, Milk} 3
Itemset
{Bread,Diaper,Milk}
Itemset Count
{Beer,Diaper} 3
{Bread,Diaper} 3
{Bread,Milk} 3 {Bread,Diaper} 
{Diaper, Milk} 3
{Bread,Milk} 
{Diaper, Milk} 
Generate Candidates Ck+1
• Are we done? Are all the candidates valid?

Item 1 Item 2 Item 3

1 2 3
1 2 3 5
1 2 5
1 4 5
Is this a valid candidate?

No. Subsets (1,3,5) and (2,3,5) should also be frequent

Apriori principle
• Pruning step:
• For each candidate (k+1)-itemset create all subset k-itemsets
• Remove a candidate if it contains a subset k-itemset that is
not frequent
 Example I
{a,b,c} {a,b,d}
• L3={abc, abd, acd, ace, bcd}
{a,b,c,d}
• Self-joining: L3*L3
– abcd from abc and abd abc abd acd bcd
– acde from acd and ace    

• Pruning:
{a,c,d} {a,c,e}
– abcd is kept since all subset itemsets are
in L3 {a,c,d,e}

– acde is removed because ade is not in L3

acd ace ade cde
• C4={abcd}   X
Generate Candidates Ck+1
• We have all frequent k-itemsets Lk
• Step 1: self-join Lk
• Create set Ck+1 by joining frequent k-itemsets that
share the first k-1 items
• Step 2: prune
• Remove from Ck+1 the itemsets that contain a subset
k-itemset that is not frequent
Computing Frequent Itemsets
• Given the set of candidate itemsets Ck, we need to compute
the support and find the frequent itemsets Lk.
• Scan the data, and use a hash structure to keep a counter
for each candidate itemset that appears in the data

Transactions Hash Structure

Ck
TID Items
1 Bread, Milk
2 Bread, Diaper, Beer, Eggs
N 3 Milk, Diaper, Beer, Coke k
4 Bread, Milk, Diaper, Beer
5 Bread, Milk, Diaper, Coke
Buckets
A simple hash structure
• Create a dictionary (hash table) that stores the
candidate itemsets as keys, and the number of
appearances as the value.
• Increment the counter for each itemset that you
see in the
 Key Value
Example {3 6 7} 0
{3 4 5} 1
{1 3 6} 3
Suppose you have 15 candidate
itemsets of length 3: {1 4 5} 5
{1 4 5}, {1 2 4}, {4 5 7}, {1 2 5}, {4 5 8}, {2 3 4} 2
{1 5 9} 1
{1 5 9}, {1 3 6}, {2 3 4}, {5 6 7}, {3 4 5},
{3 6 8} 0
{3 5 6}, {3 5 7}, {6 8 9}, {3 6 7}, {3 6 8}
{4 5 7} 2
{6 8 9} 0
Hash table stores the counts of the {5 6 7} 3
candidate itemsets as they have been
computed so far {1 2 4} 8
{3 5 7} 1
{1 2 5} 0
{3 5 6} 1
{4 5 8} 0
 Subset Generation
Given a transaction t, what Transaction, t
are the possible subsets of
1 2 3 5 6
size 3?

Level 1
1 2 3 5 6 2 3 5 6 3 5 6

Level 2

12 3 5 6 13 5 6 15 6 23 5 6 25 6 35 6

123
135 235
Recursion! 125
136
156
236
256 356
126

Level 3 Subsets of 3 items

 Key Value
Example {3 6 7} 0
{3 4 5} 1
{1 3 6} 3
Tuple {1,2,3,5,6} generates the
following itemsets of length 3: {1 4 5} 5
{2 3 4} 2
{1 5 9} 1
{1 2 3}, {1 2 5}, {1 2 6}, {1 3 5}, {1 3 6},
{3 6 8} 0
{1 5 6}, {2 3 5}, {2 3 6}, {3 5 6},
{4 5 7} 2
{6 8 9} 0
Increment the counters for the itemsets {5 6 7} 3
in the dictionary
{1 2 4} 8
{3 5 7} 1
{1 2 5} 0
{3 5 6} 1
{4 5 8} 0
 Key Value
Example {3 6 7} 0
{3 4 5} 1
{1 3 6} 4
Tuple {1,2,3,5,6} generates the
following itemsets of length 3: {1 4 5} 5
{2 3 4} 2
{1 5 9} 1
{1 2 3}, {1 2 5}, {1 2 6}, {1 3 5}, {1 3 6},
{3 6 8} 0
{1 5 6}, {2 3 5}, {2 3 6}, {3 5 6},
{4 5 7} 2
{6 8 9} 0
Increment the counters for the itemsets {5 6 7} 3
in the dictionary
{1 2 4} 8
{3 5 7} 1
{1 2 5} 1
{3 5 6} 2
{4 5 8} 0
The Hash Tree Structure
Suppose you have the same 15 candidate itemsets of length 3:
{1 4 5}, {1 2 4}, {4 5 7}, {1 2 5}, {4 5 8}, {1 5 9}, {1 3 6}, {2 3 4},
{5 6 7}, {3 4 5}, {3 5 6}, {3 5 7}, {6 8 9}, {3 6 7}, {3 6 8}
You need:
• Hash function
• Leafs: Store the itemsets

Hash function = x mod 3 234

3,6,9 567
1,4,7 145 356 367
345
2,5,8 136 368
357
124 689
457 125 159
458

At the i-th level we hash on the i-th item

Subset Operation Using Hash Tree
Hash Function
1 2 3 5 6 transaction

1+ 2356
2+ 356 1,4,7 3,6,9
2,5,8
3+ 56

234
567

145 136
345 356 367
357 368
124 159 689
125
457 458
 Subset Operation Using Hash Tree
Hash Function
1 2 3 5 6 transaction

1+ 2356
2+ 356 1,4,7 3,6,9
12+ 356
2,5,8
3+ 56
13+ 56
234
15+ 6 567

145 136
345 356 367
357 368
124 159 689
125
457 458
 Subset Operation Using Hash Tree
Hash Function
1 2 3 5 6 transaction

1+ 2356
2+ 356 1,4,7 3,6,9
12+ 356
2,5,8
3+ 56
13+ 56
234
15+ 6 567 Increment the counters

145 136
345 356 367
357 368
124 159 689
125
457 458 Match transaction against 9 out of 15 candidates

Hash-tree enables to enumerate itemsets in transaction

and match them against candidates
 Count All pairs Count
All the items of items the pairs
items from L1

C1 Filter L1 Construct C2 Filter L2 Construct C3

First Second
pass pass

Frequent Frequent
items pairs
 43

APriori for All Frequent Itemsets

• One pass for each k.
• Needs room in main memory to count each
candidate k -set.
• For typical market-basket data and reasonable
support (e.g., 1%), k = 2 requires the most
memory.
 44

Picture of APriori

Item counts Frequent items

Counts of
pairs of
frequent
items

Pass 1 Pass 2
 45

Details of Main-Memory Counting

• Two approaches:
1. Count all pairs, using a “triangular matrix” = one
dimensional array that stores the lower diagonal.
2. Keep a table of triples [i, j, c] = “the count of the
pair of items {i, j } is c.”
• (1) requires only 4 bytes/pair.
• Note: always assume integers are 4 bytes.
• (2) requires 12 bytes, but only for those pairs
with count > 0.
 46

12 per
4 per pair
occurring pair

Method (1) Method (2)

 47

Triangular-Matrix Approach
• Number items 1, 2,…
• Requires table of size O(n) to convert item names to
consecutive integers.
• Count {i, j } only if i < j.
• Keep pairs in the order {1,2}, {1,3},…, {1,n },
{2,3}, {2,4},…,{2,n }, {3,4},…, {3,n },…{n -1,n }.
 48

Triangular-Matrix Approach
• Find pair {i, j } at the position
(i –1)(n –i /2) + j – i.

• Total number of pairs n (n –1)/2; total bytes

about 2n 2.
 49

Details of Approach #2

• Total bytes used is about 12p, where p is the

number of pairs that actually occur.

•
ASSOCIATION RULES
 Association Rule Mining
• Given a set of transactions, find rules that will predict the
occurrence of an itemset based on the occurrence of other
itemset in the transaction

Market-Basket transactions
Example of Association Rules
TID Items
{Diaper}  {Beer},
1 Bread, Milk {Milk, Bread}  {Eggs,Coke},
2 Bread, Diaper, Beer, Eggs {Beer, Bread}  {Milk},
3 Milk, Diaper, Beer, Coke
4 Bread, Milk, Diaper, Beer Implication means co-occurrence,
not causality!
5 Bread, Milk, Diaper, Coke
 Definition: Association Rule
 Association Rule
TID Items
– An implication expression of the form
X  Y, where X and Y are itemsets 1 Bread, Milk
– Example: 2 Bread, Diaper, Beer, Eggs
{Milk, Diaper}  {Beer} 3 Milk, Diaper, Beer, Coke
 Rule Evaluation Metrics 4 Bread, Milk, Diaper, Beer
5 Bread, Milk, Diaper, Coke
– Support (s)
 Fraction of transactions that contain Example:
both X and Y
 the probability P(X,Y) that X and Y
{Milk , Diaper }  Beer
occur together
 (Milk, Diaper, Beer ) 2
– Confidence (c) s   0.4
 Measures how often items in Y |T| 5
appear in transactions that
 (Milk, Diaper, Beer ) 2
contain X c   0.67
 the conditional probability P(Y|X) that Y  (Milk, Diaper ) 3
occurs given that X has occurred.
Association Rule Mining Task
• Input: A set of transactions T, over a set of items I
• Output: All rules with items in I having
• support ≥ minsup threshold
• confidence ≥ minconf threshold
Mining Association Rules
• Two-step approach:
1. Frequent Itemset Generation
– Generate all itemsets whose support  minsup

2. Rule Generation
– Generate high confidence rules from each frequent itemset,
where each rule is a partitioning of a frequent itemset into
Left-Hand-Side (LHS) and Right-Hand-Side (RHS)

Frequent itemset: {A,B,C,D}

Rule: ABCD
Rule Generation
• We have all frequent itemsets, how do we get the
rules?
• For every frequent itemset S, we find rules of the form
L  S – L , where L  S, that satisfy the minimum confidence
requirement
• Example: L = {A,B,C,D}
• Candidate rules:
A BCD, B ACD, C ABD, D ABC
AB CD, AC  BD, AD  BC, BD AC, CD AB,
ABC D, BCD A, BC AD,
• If |L| = k, then there are 2k – 2 candidate association
rules (ignoring L   and   L)
Rule Generation
• How to efficiently generate rules from frequent
itemsets?
• In general, confidence does not have an anti-monotone
property
c(ABC D) can be larger or smaller than c(AB D)

• But confidence of rules generated from the same

itemset has an anti-monotone property
• e.g., L = {A,B,C,D}:

c(ABC  D)  c(AB  CD)  c(A  BCD)

• Confidence is anti-monotone w.r.t. number of items on the RHS

of the rule
 Rule Generation for Apriori Algorithm
ABCD=>{ } }
ABCD=>{
Low
Confidence
Rule
BCD=>A
BCD=>A ACD=>B
ACD=>B ABD=>C
ABD=>C ABC=>D
ABC=>D

CD=>AB
CD=>AB BD=>AC
BD=>AC BC=>AD
BC=>AD AD=>BC
AD=>BC AC=>BD
AC=>BD AB=>CD
AB=>CD

D=>ABC
D=>ABC C=>ABD
C=>ABD B=>ACD
B=>ACD A=>BCD
A=>BCD
Pruned
Rules
Lattice of rules created by the RHS
Rule Generation for APriori Algorithm
• Candidate rule is generated by merging two rules that
share the same prefix
in the RHS
CD->AB BD->AC

• join(CDAB,BDAC)
would produce the candidate
rule D  ABC

• Prune rule D  ABC if its

subset ADBC does not have D->ABC
high confidence

• Essentially we are doing APriori on the RHS

Thực hành
• http://www.philippe-fournier-
viger.com/spmf/Apriori.php