ASSOCIATION RULE MINING BASED ON TRADE LIST

International Journal of Data Mining & Knowledge Management Process (IJDKP) Vol.1, No.6, November 2011
DOI : 10.5121/ijdkp.2011.1603 25
ASSOCIATION RULE MINING BASED
ON TRADE LIST
Ms. Sanober Shaikh1
Ms. Madhuri Rao2
1
Department of Information Technology, TSEC, Bandra (w), Mumbai
s.sanober1@gmail.com
2
Department of Information Technology, TSEC, Bandra (w), Mumbai
my_rao@yahoo.com
ABSTRACT
In this paper a new mining algorithm is defined based on frequent item set. Apriori Algorithm scans the
database every time when it finds the frequent item set so it is very time consuming and at each step it
generates candidate item set. So for large databases it takes lots of space to store candidate item set .In
undirected item set graph, it is improvement on apriori but it takes time and space for tree generation.
The defined algorithm scans the database at the start only once and then from that scanned data base it
generates the Trade List. It contains the information of whole database. By considering minimum support
it finds the frequent item set and by considering the minimum confidence it generates the association rule.
If database and minimum support is changed, the new algorithm finds the new frequent items by scanning
Trade List. That is why it’s executing efficiency is improved distinctly compared to traditional algorithm.
KEYWORDS
Undirected Item set Graph, Trade List
1. INTRODUCTION
Mining Association rule is very important field of research in data mining. The problem of
mining Association rule is put forward by R.S Agarwal first in 1993. Now the Association rules
are widely applied in E-commerce, bank credit, shopping cart analysis, market analysis, fraud
detection, and customer retention, to production control and science exploration. etc. [1]
Now a days we will find many mining methods for finding the frequent item set such as Apriori
algorithm, Frequent Pattern-Tree algorithm etc. Apriori algorithm’s disadvantage is it generates
lot of candidate itemsets and scans database every time. If database contains huge number of
transactions then scanning the database for finding the frequent itemset will be too costly and it
generates a lot of candidates. Next FP-Tree algorithm’s advantage is it does not produce any
candidate items but it scans database two times in the memory allowed. But when the memory
does not meet the need, this algorithm becomes more complex. It scans the database more than
two times and the I/O expenses will increase [2]. That is why there is need to design an efficient
algorithm which updates, protects and manages the association rule in large transactional
database. So far many researchers made analysis and research for how to efficiently update the
association rules and put forward corresponding algorithm. There are two instances in the
problem of Association Rule update. The first instance is when the database is changed then
how to find frequent item sets. FUFIA Algorithm is the representational updating method for
this problem. The second instance is when the minimum support is changed then how to find
frequent items sets. IUA algorithm is the representational updating method for this problem.
These updating algorithms have both advantages and disadvantages. This paper proposes a
dynamic algorithm of frequent mining based on undirected item set graph which scans the
database only once and then saves the information of original database in undirected item set

26
graph and finds the frequent item sets directly from the graph. It does not generate any
candidate items. When database and minimum support is changed, the algorithm rescans the
undirected item set graph to get the new frequent item sets.[3]
2. BASIC CONCEPT OF ASSOCIATION RULE
Association rule finds interesting associations and/or correlation relationships among large set
of data items. Association rule shows attribute value conditions that occur frequently together
in a given dataset. A typical and widely-used example of association rule mining is Market
Basket Analysis.
For example, data are collected using bar-code scanners in supermarket. Such ‘market basket’
databases consist of a large number of transaction records. Each record lists all items bought by
a customer on a single purchase transaction. Managers would be interested to know if certain
groups of items are consistently purchased together. They could use this data for adjusting store
layouts (placing items optimally with respect to each other), for cross-selling, for promotions,
for catalog design and to identify customer segments based on buying patterns.
Association rules do not represent any sort of causality or correlation between the two item sets
The problem of mining association rules can be described as below: if I = {I1, I2 ….In} is the
set of items. Suppose D is database transaction set and each transaction T contains set of items,
such that T⊆I. Each transaction has identifier called as TID i.e. transaction id. Suppose A is a
set of items and transaction T is said to contain A only if A ⊆T.
Association rule is an implication like as A ⇒B in which A, B ⊂ I and A∩B=∅. [6]
Definition of support: The support is the percentage of transactions that demonstrate the rule. An
item set is called frequent if its support is equal or greater than an agreed upon minimal value –
the support threshold. [8]
Definition of Confidence: Every association rule has a support and a confidence.
An association rule is of the form: X => Y.
X => Y: if someone buys X, he also buys Y.
The confidence is the conditional probability that, given X present in a transition, Y will also be
present. Confidence measure, by definition:
Confidence(X=>Y) = support(X, Y)/ support(X)
The aim of association rule is to find all association problems having support and confidence
not less than given threshold value. For the given support i.e. minsupp, if the item set of D’s
support is not less than minsupp, then it can say that D is the frequent item set.
3. FINDING FREQUENT ITEM SETS
First step is to scan the database. It makes each item as a node and at the same time it makes
the supporting trade list for each node. Supporting trade list is a binary group T= {Tid, Itemset}
(where Tid is transaction id and Itemset is trade item set). Given database that includes five
items and nine transactions (shown in table one). Suppose that minimum support minsupp is
two. Table two contains the information of support trade list of table one.
With this Trade List directly we will get information of which items are appearing in which
transactions. So here number of transactions related to that item will decide count of that item.
So we have count of I1 as 6 as shown in Table 2. Similarly we will get the count of all the
items in the database. Now after considering the minimum support from user we will compare
that minimum support with the count. If it is greater those will be considered as frequent-1 item
set.

27
Table 1: A Store Business Data
TID The List Of Item ID
T100 I1,I2,I5
T200 I2,I4
T300 I2,I3
T400 I1,I2,I4
T500 I1,I3
T600 I2,I3
T700 I1,I3
T800 I1,I2,I3,I5
T900 I1,I2,I3
Table 2: Trade List of Commodity Item
Commodity
Item
Support Trade List
I1 T100,T400,T500,T700,
T800, T900
I2 T100,T200,T300,T400,
T600,T800, T900
,T800,T9
I3 T300,T500,T600,T700,
T800,T900
,T
I4 T200,T400
I5 T100,T800
In next step for finding frequent itemset do intersection of I1 and I2. In result if we will get
some transactions we will get common then it means that the item is related to other transaction
also. Count the numbers of those common transactions that will give the count of those two
items that are bought together that many numbers of times. Example I1∩ I2 will get the count
as 4 that means I1 and I2 are together 4 number of times in the database. Compare this with
minimum support. Then we will get frequent-2 itemset. Similarly the procedure is iteratively
applied.
3.1. Updating Trade List
When database and minimum support i.e. minsupp is changed the Trade List should be
changed accordingly. If we want to add some new items to the database, then Trade List is
updated accordingly.
3.2.1. Database Affair Changed
For example, when a new item T910 is added to table one; the result is as shown as in table
three.

28
Table 3: The New Data in a Store
TID The list of items
T100 I1,I2,I5
T200 I2,I4
T300 I2,I3
T400 I1,I2,I4
T500 I1,I3
T600 I2,I3
T700 I1,I3
T800 I1,I2,I3,I5
T900 I1,I2,I3
T910 I1,I4
A new item T910 have added at this time. So the arisen number of side <I1, I4> is two. As
shown in fig.1, frequent 1-item set is L1= {I1, I2, I3, I4, I5};
frequent 2-item set is L2={{ I1 , I2},{ I1 , I3 },{ I1 , I5},{ I2 , I3 },{ I2 , I4 },
{I2, I5}, {I1, I4 }}; frequent 3-item set is L3={{ I1, I2, I3 },{ I1, I2, I5},{ I1, I2, I4}}.
3.2.2 Minimum support changed
For example, when the minimum support minsupp is three, frequent 1-item set={I1, I2 , I3 };
frequent 2- item is L2={{ I1, I2},{ I1 , I3},{ I2 , I3 }}.
4. RESULTS
4.1 Results of Apriori Algorithm
Fig1: Frequent Item Set with Apriori Algorithm with database shown in Table 1

29
4.2 Results of Trade List
Fig 2: Main Form
In Fig 2 form the first i.e. Item Set File asks for the database from which you want to retrieve
the frequent items. Here for input of Item set file one .isf file is made as shown in Fig 3. In that
file the code for connectivity with database is made. Through the code the database is converted
to a text file. In the first line write name of .isf file that will be converted to a format which the
code will accept.
Fig 3: Item Set File

30
When we will click on Generate button in Fig 2, Trade list is made from which we can come to
know how many number of items are present in input database as shown in Fig 4.
Fig 4: Trade List
Then with the help of this Trade list we will get frequent items easily.Here minimum support is
3. Now the count of each item is compared with minimum support. If count is greater than
minimum support those items will be frequent item sets as shown in fig 5.

31
Fig 5: Frequent Items
Fig 5 (cont): Frequent Items

32
Fig 5 (cont): Frequent Items
Confidence of each item is compared with minimum confidence given by user and strong
association rule is formed. The items having confidence greater than or equal to minimum
confidence, are stored in file shown in Fig 6.
Fig 6: Association Rule

33
5. CONCLUSION
In this project candidate items are not generated. The information of items of original database
is saved in undirected item set graph. Then the information of frequent item set is found by
searching trade list. The apriori scans the database too many times and generating candidates in
each step. If we have huge amount of data then scanning such data and storage of huge amount
of candidates is very difficult. Algorithm based on “A new association rule mining based on
undirected itemset graph” having the disadvantage of tree generation. It takes time for
generating tree. Now Trade list technique as compare to apriori and undirected itemset graph
takes less amount of time and give the proper results.
6. ACKNOWLEDGEMENT
1. Ms Madhuri Rao(Guide)
2. Mr. Naushad Shaikh
7. REFERENCES
[1] S. Chai, J. Yang, Y. Cheng, “The Research of Improved Apriori Algorithm for Mining
Association Rules”, 2007 IEEE.
[2] S. Chai, H. Wang, J. Qiu, “DFR: A New Improved Algorithm for Mining Frequent Item
sets”, Fourth International Conference on Fuzzy Systems and Knowledge Discovery
(FSKD 2007)
[3] R. Agrawal, T. Imielinski, A. Swami, “Mining Association Rules between Sets of Items
in Very Large Databases [C]”, Proceedings of the ACM SIGMOD Conference on
Management of Data, Washington, USA, 1993-05: 207-216
[4] R. Agrawal, T. Srikant, “Fast Algorithms for Mining Association Rules in Large Database
[C]”, Proceedings of 20th VLDB Conference, Santiago, Chile, 1994: 487-499
[5] L Guan, S Cheng, and R Zhou, “Mining Frequent Patterns without Candidate Generation
[C]”, Proceedings of SIGMOD’00, Dallas, 2000:1-12.
[6] Dongme Sun, Shaohua Teng, Wei Zhang, “An algorithm to improve the effectiveness of
Apriori”, Proceedings of 6th IEEE International Conference on Cognitive Informatics
(ICCI'07), IEEE2007.
[7] http://en.wikipedia.org/wiki/Apriori_algorithm.

ASSOCIATION RULE MINING BASED ON TRADE LIST

More Related Content

Similar to ASSOCIATION RULE MINING BASED ON TRADE LIST (20)

Recently uploaded (20)

ASSOCIATION RULE MINING BASED ON TRADE LIST