International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 5

ISSN: 2321-8169
1205 1208

_______________________________________________________________________________________________

Design and Implementation of Efficient APRIORI Algorithm

Rupinder Kaur, Rajeev Bedi, S. K. Gupta
Abstract: - Apriori algorithm is the classical algorithm used for association rule mining. This paper presents out the overview of basic
approaches used with the classical Apriori algorithm and formulates the problems associated with the classical approaches. This paper presents
out an enhanced Apriori algorithm that overcomes these limitations and is more efficient.

__________________________________________________*****_________________________________________________
I.

INTRODUCTION

Association rule mining is one of the data mining tasks that

are used for finding correlations between the transactions.
Association rules are if/then statements that are used for
finding relations between data in the database. An
association rule has two parts, an antecedent (if) and a
consequent (then). An antecedent is an item that is found in
the data. A consequent is an item that is found in
combination with the antecedent. Association rules are
created by analyzing data for frequent if/then patterns and
then support and confidence are the parameters that are used
to identify the most important relationships. Learning
association rules basically means finding the items that are
purchased together more frequently than others. Shopping
centers use association rules to place the items next to each
other so that users buy more items. Apriori is the classic and
probably the most basic algorithm to do association rule
mining.
II.

ASSOCIATION RULE MINING ALGORITHMS

Classical apriori algorithm

Apriori algorithmic rule is basic algorithmic rule for
association rule mining. It takings by distinctive the frequent
individual things within the data and lengthening them to
larger and bigger item sets as long as those item sets seem
sufficiently usually within the data. The frequent item sets
verified by Apriori are often used to determine association
rules that highlight general trends within the data.
Apriori uses a bottom-up approach, wherever frequent
subsets are extended one item at a time( a step called
candidate generation ), and tested against the data.
Algorithmic rule terminates once no winning extension units
are found. Apriori algorithmic rule generates frequent item
sets. If association item satisfies a definite minimum support
and minimum confidence then its thought about as a
frequent item. This whole algorithmic rule relies on plan of
looking out level by level.
Association rule mining is a 2 step process:i) Find all the frequent item sets from the data. If support of
associate item set A is larger than the minimum support i.e.,
support(A)>=minsup, them itemset a is thought as frequent
itemset otherwise not a frequent itemset.
ii) Generate association rules from the frequent itemsets.

Improved Apriori algorithm

Huiyang wang etc.al proposed two theorems to improve the
Aprirori algorithm to reduce the times of scanning
frequency itemsets.
Theorem1.:- suppose X and Y are two subsets of transaction
T and X is subset of Y. if Y is frequent itemset then X must
be frequent itemset.
Theorem 2:- suppose X and Y are two subsets of transaction
T and X is subset of Y. if Y is not frequent itemset then X
must not be frequent itemset.
Weighted apriori algorithm
Weighted approach with the basic APRIORI was introduced
to address the problem of using single minimum support for
selecting the frequent item sets. In the transactional
databases items are not uniformly distributed. Use of single
minimum support lead to either missing of rare association
rules if set too high or lead to combination explosion if set
too low. Weighted association rules deal with this issue. To
reflect different importance to different items, weights were
assigned to different items.
Consider D- transaction database
I= {i1, i2, i3} = set of items. Each transaction is
subset of I with transaction id-TID.
Then W= {w1,
corresponding to I.

w2,

w3.} is the

weight set

Classical algorithm was first used to obtain the frequent

item sets without weights. After weight assigning approach,
attributes with weighted support less than minimum
weighted support were removed.
PROBLEMS
ASSOCIATED
ALGORITHM
1.
2.

WITH

APRIORI

Candidate generation tries to load maximum no. of

subsets before each scan increasing execution time.
Bottom-up approach increases no. of scans required
for maximal subset.
III.
PROPOSED ALGOTIHM

CSk: Candidate item set of size k

LSk: Set of frequent items of size k
LS1= {frequent items};
Sort Item set LSk.
For (k= 1; Lk!=; k++) do begin
1205

IJRITCC | May 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 5

ISSN: 2321-8169
1205 1208

_______________________________________________________________________________________________
CSk+1= candidates generated from LSk;

8. If No. Of Rules >= required No. Of rules Exit.

For each transaction t in database do

9. List Rules.

Increment count for items in CSk+1

LSk+1= candidates in Ck+1with min_support

Enhanced Apriori algorithm scans data base once. For each

row it builds list of possible pairs/permutations of elements.
On next row these build pairs are evaluated. After all
scanning rules are generated and disqualifying pair,
Elements are discarded from results.

End

HOW THE PROBLEM IS SOLVED

Return kLk;

1. Managing candidate items using sorted list reduces time

required to scans Items. Items are maintained in sorted form
so it requires lesser amount of time to insert new candidate
item.

That is contained in t
Sort candidate set CSk.

Steps:
1. Initialize variables
2. For each transaction in the DB T repeat 3-7:
3. Processor scans DB
identification set (TID).

and

creates

the

transaction

4. Apply Weights to Item sets.

2. Breadth first each: BFS helps in finding building rules as

we scan database instead of repeatedly scan database when
building rules. Because as we scan each transaction:
associations are generated.

5. For Each Row Repeat

IV.

a)

Prepare pairs common in rows.

b)

Eliminate pairs without common elements except

last element.

6. Calculate candidate k-item set counts, when the count is

greater than s, let freqk be frequent k-item sets.

EXPERIMENTAL RESULTS

Enhanced Apriori algorithm used the hybrid approach for

the association rule mining. Enhanced apriori algorithm
made use of concept of soting with the weighted approsch to
mine the association rules efficiently with the minimum
time possible.

7. Sort candidate Item sets.

Figure 1:- output screen

1206
IJRITCC | May 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 5

ISSN: 2321-8169
1205 1208

_______________________________________________________________________________________________
Choose confidence 90% and 70% and analyze the curves of
time change with the change in support. As the support

increases runtime reduces gradually that shows the stability

of the proposed algorithm.

Figure 2:- Time change curve 1

Figure 3:-time change curve 2

Figure 4:- Reduced association rule records

1207
IJRITCC | May 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 5

ISSN: 2321-8169
1205 1208

_______________________________________________________________________________________________
Figure 4 shows mining of database with association rule
with different minimum support degree. Horizontal axle
shows different support degrees in percentage and vertical
axle shows number of reducing records in the same
database.
REFERENCES
[1] Chun-Hao Chen, Guo-Cheng Lan, Tzung-Pei Hong, and YuiKai Lin, A High Coherent Association Rule Mining
Algorithm in the proceedings of IEEE international Conference
on Technologies and Applications of Artificial Intelligence,
Nov. 2012 , pp.1 4.
[2] Chun Zhang, Dezan Xie, Ning Zhang, HonghuiLi, The
Improvement of Apriori Algorithm and Its Application in
Fault Analysis of CRH EMU in the proceedings of IEEE
international conference on Service Operations, Logistics, and
Informatics (SOLI) , July 20l1, pp. 543 547.
[3] Hangbin LI, Shuhua CHEN, Jianchen LI, Shuo WANG,
Yihang FU An Improved Multi-Support Apriori Algorithm
Under the Fuzzy Item Association Condition in the
proceedings of IEEE international conference on Electronics,
Communications and Control (ICECC) Sept 2011, pp: 35393542.
[4] Huiying Wang, Xiangwei Liu, The Research of Improved
Association Rules Mining Apriori Algorithm in the
proceedings of IEEE Eighth International Conference on Fuzzy
Systems and Knowledge Discovery, July 2011, pp. 961-964.
[5] Idheba Mohamad Ali O. Swesi, Azuraliza Abu Bakar, Anis
Suhailis Abdul Kadir, Mining Positive and Negative
Association Rules from Interesting Frequent and Infrequent
Itemsets in the proceedings of IEEE 9th International Conference
on Fuzzy Systems and Knowledge Discovery (FSKD), May
2012, pp: 650-655.
[6] Jia Baohui,Wang Yuxin, Yang Zheng-qing, The Research of
Data Mining in AHM Technology based on Association Rule
in the proceedings of IEEE conference on Prognostics & System
Health Management May 2011, pp:1-8.
[7] Lei Chen, The Research of Data Mining Algorithm Based
on Association Rules in the proceedings of ICCASM 2nd
International Conference on Computer Application and System
modeling, 2012, pp: 0548-0551.
[8] Luo Fang, Qiu Qizhi, The Study on the Application of Data
Mining Based on Association Rules in the proceedings of IEEE
International Conference on Communication Systems and
Network Technologies May 2012, pp: 477-480.

Space Complexity in the proceedings of IEEE annual Indian

conference (INDICON) Dec 2012, pp: 1105-1110.
[10] Qiang Yang, Yanhong Hu, Application of Improved
Apriori Algorithm on Educational Information in the
proceedings of IEEE Fifth International Conference on Genetic
and Evolutionary Computing, Sept 2011, pp:330-332.
[11] Rina Raval, Prof. Indr Jeet Rajput , Prof. Vinitkumar Gupta,
Survey on several improved Apriori algorithm IOSR Journal
of Computer Engineering (IOSR-JCE) e-ISSN: 2278-0661, pISSN: 2278-8727 Volume 9, Issue 4 (Mar. - Apr. 2013), pp: 57-61.
[12] Rui Chang, Zhiyi Liu ,An Improved Apriori Algorithm in
the proceedings of IEEE International Conference on Electronics
and Optoelectronics (ICEOE), July 2011,pp: 476-478.
[13] Rupinder Kaur, Rajeev kumar Bedi, Sunil Kumar Gupta,
Review of association rule mining, International Journal of
Advanced Technology & Engineering Research (IJATER), ISSN
No: 2250-3536, Volume 4, Issue 2, March 2014,pp:14-17.
[14] Sandeep Singh Rawat, Lakshmi Rajamani, Probability
Apriori based Approach to Mine Rare Association Rules in
the proceedings of IEEE 3rd Conference on Data Mining and
Optimization (DMO) June 2011, pp: 253-258.
[15] S.Suriya , Dr.S.P.Shantharajah , R.Deepalakshmi , A
Complete Survey on Association Rule Mining with Relevance
to Different Domain International journal of advanced scientific
and technical research issue2, volume1 , Feb 2012, pp:163-168.
[16] Suraj P. Patil, A Novel Approach for Efficient Mining and
Hiding of Sensitive Association Rule in the proceedings of
Nirma university international conference on engineering,
December 2012, pp:1-6.
[17] Suraj P . Patil, U. M. Patil and Sonali Borse, The novel
approach for improving apriori algorithm for mining
association Rule World Journal of Science and Technology
2012, pp:75-78 .
[18] Xu Chil, ZHANG Wen Fang, Review of Association Rule
Mining Algorithm in Data Mining in the proceedings of IEEE
3rd International Conference on Communication Software and
Networks May 2011, pp. 512-516.
[19] Zhuang Chen, Shibang Cai, Qiulin Song and Chonglai Zhu
,An Improved Apriori Algorithm Based on Pruning
Optimization and Transaction Reduction in the proceedings of
IEEE 2nd international conference on Artificial Intelligence,
Management Science and Electronic Commerce (AIMSEC)
Aug 2011, pp: 1908-1911.

[9] Punit Mundra, Amit K Maurya, and Sanjay Singh, Enhanced

Mining Association Rule Algorithm with Reduced Time &

1208
IJRITCC | May 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________