Mining Maximum Frequent Item Sets Over Data Streams Using Transaction Sliding Window Techniques

International Journal of Information Technology Convergence and Services (IJITCS) Vol.3, No.2, April 2013
DOI:10.5121/ijitcs.2013.3201 1
Mining Maximum Frequent Item Sets Over Data
Streams Using Transaction Sliding
Window Techniques
Neeraj And Anuradha
Student/ Software engineering/M.Tech ITM University Gurgaon, 122017, India
neeraj.yadav28 @gmail.com
Faculty/ Software engineering ITM University Gurgaon, 122017, India
anuradha@itmindia.edu
ABSTRACT
As we know that the online mining of streaming data is one of the most important issues in data mining. In
this paper, we proposed an efficient one- .frequent item sets over a transaction-sensitive sliding window),
to mine the set of all frequent item sets in data streams with a transaction-sensitive sliding window. An
effective bit-sequence representation of items is used in the proposed algorithm to reduce the time and
memory needed to slide the windows. The experiments show that the proposed algorithm not only attain
highly accurate mining results, but also the performance significant faster and consume less memory than
existing algorithms for mining frequent item sets over recent data streams. In this paper our theoretical
analysis and experimental studies show that the proposed algorithm is efficient and scalable and perform
better for mining the set of all maximum frequent item sets over the entire history of the data streams.
Keywords
Recently frequent item sets, Maximum frequent item set, Transaction sliding window, Data stream.
1. INTRODUCTION
We know that the frequent item sets mining play an essential role in many data mining tasks.
With years of research into this research, several data stream mining problems have been
discussed, such as frequent item set mining [5, 6, 7, 8, 9], closed frequent structure mining
[19,24,29]. The frequent item set mining over data streams is to find an approximate set of
frequent item sets in transaction with respect to a given support and threshold. It should support
the flexible trade-off between processing time and mining accuracy. It should be time efficient
even when the user-specified minimum support threshold is small. The objective was to propose
an effective algorithm which generates frequent patterns in a very less time. Our approach has
been developed based on improvement and analysis of MFI algorithm. Frequent patterns can be
very meaningful in data streams. In network monitoring, frequent patterns can correspond to
excessive traffic which could be an indicator for network attack. In sales transactions, frequent
patterns relate to the top selling products in a market, and possibly their relationships. Due to
numerous applications of this sort mining frequent patterns from data stream has been a hot
research area. If we consider that the data stream consists of transactions, each being a set of
items, then the problem definition of mining frequent patterns can be written as - Given a set of
transactions, find all patterns with frequency above a threshold s. Traditional mining algorithms
assume a finite dataset over which the algorithms are allowed to scan multiple times. Therefore it
is necessary to modify these algorithms in order to apply these algorithms to transactional data
streams. An important property that distinguishes data stream mining from traditional data mining

2
is the unbounded nature of stream data, which precludes multiple-scan algorithms. Traditional
frequent item sets mining algorithms are thus not applicable to a data stream [11].
Stream mining algorithms are being developed to discover useful knowledge from data as it
streams in -- a process that isn't as straightforward as it might seem. Scientists have to deal with
the fact that the data rate of the stream isn't constant, leading to a condition called business, and
the patterns of the data stream are continuously evolving. Online mining of frequent item sets
over a stream sliding window is one of the most important problems in stream data mining with
broad applications. It is also a difficult issue since the streaming data possess some challenging
characteristics, such as unknown or unbound size, possibly a very fast arrival rate, inability to
backtrack over previously arrived transactions, and a lack of system control
overtheorderinwhichthedataarrive.Inthispaper,weproposeaneffectivebit-sequencebased,one pass
algorithm, called MFI-TransSW (Mining Frequent Item sets within a Transaction-sensitive
Sliding Window), to mine the set of frequent item sets from data streams within a transaction-
sensitive sliding window which consists of a fixed number of transactions[2,28,30]
Figure first shows the categories of data streams are divided into three parts: Landmark windows,
sliding windows, damped windows. Further the sliding windows also divided into two more parts:
transaction sensitive windows and time sensitive sliding windows [2, 21, 22, 23].
Fig 1.Categories of data stream.
Pei, Yin, & Mao, 2004 is used to maintain the frequent item sets and their supports stored in
titled-time windows. In a sliding window model, knowledge discovery is performed over a fixed
number of recently generated data elements which is the target of data mining. Two types of
sliding widow, i.e., transaction-sensitive sliding window and time-sensitive sliding window, are
used in mining data streams. The basic processing unit of window sliding of transaction-sensitive
sliding window is an expired transaction while the basic unit of window sliding of time-sensitive
sliding window is a time unit, such as a minute or 1h. [3, 20, 25, 26, 27]
According to the stream processing model, the research of mining frequent item sets in data
streams can be divided into three categories, snapshot windows, sliding windows and Landmark
windows. In snapshot window model the frequent items are detected in a fixed range of time so
that model is not used frequently. In landmark window model knowledge discovery is performed
based on the values between the specific timestamp called landmark and the present. In the
sliding window model knowledge discovery is performed over a fixed number of recently
generated data elements which is the target of data mining. Sliding window model is a widely

3
used model to perform frequent item set mining since it considers only recent transactions and
forgets obsolete ones. Due to this reason, a large number of sliding window based algorithms
have been devised [11][12] [13][14][15][17]. However, only a few of these studies adaptively
maintain and update the set of frequent item sets [13][14][15] and others only store sliding
window transactions in an efficient way using a suitable synopsis structure and perform the
mining task when the user requests.
The purpose of this paper is to mining the frequent item set over online data streams with the help
of the transaction sensitive sliding window. The experiment shows that the purposed algorithm
MFI-TransSW not only attained high accurate mining data but also increase the speed as well as
the memory uses is less than the existing mining algorithm. The problem of mining frequent item
set is defined and the algorithm is purposed also in this paper.
2. PROBLEM DEFINITION
One of the most important data mining problems is mining maximal frequent item sets from a
large database [2, 6, 7, 14, 25].This problem of mining maximal frequent item sets was first of all
given by Bayardo [4].
Let W= {i1, i2,..., im} be a set of items. A transaction T= (tid, x1x2...xn), xi 2W, for is a set of
items, while n is called the size of the transaction, and tid is the unique identifier of the
transaction. An item set is a non empty set of items. An item set with size k is called a k-item set.
A transaction data stream TDS=T1,T2,...,TN is a continuous sequence of transactions, where N is
the tid of latest incoming transaction TN.A transaction-sensitive sliding window (TransSW) in the
transaction data stream is a window that slides forward for every transaction. The window at each
slide has a fixed number, w, of transactions, and w is called the size of the window. Hence, the
current transaction- sensitive sliding windows
TransSWNw+1=[TN w+1,TN w+2,...,TN],where w+1 is the window identifier of current
TransSW.The support of an item set X over TransSW, denoted as sup(X) TransSW, is the number
of transactions in TransSW containing X as a subset. An item set X is called a frequent item set
(FI) if sup(X) TranSWPs w, where s is a user defined minimum support threshold (MS) in the
range of [0, 1, 2]. The value s w is called the frequent threshold of TranSW (FTTranSW) given
a transaction-sensitive sliding window TransSW, and a MST s, the problem of online mining of
frequent item sets in recent transaction data streams is to mine the set of all frequent item sets by
one scan of the TransSW.
EXAMPLE: Suppose we have six transaction in a transaction data stream be <T1,(1 3 4
7)>,<T2,(2 3 5 6 8)>,<T3,(1 2 3 5 7) >,<T4,(2 5 8)>,<T5,(1 3 7 8)>,<T6,(2 3 7)>, where
T1,T2,T3,T4,T5,T6 are the transactions and the 1,2,3.,4,5,6,7,8 are the item sets in the
transactions .Let the size of the sliding window be 3 and the minimum support is 2.

4
TID
T1
T2
T3
T4
T5
T6
Item sets
1,3,4,7
2,3,5,6,8
1,2,3,5,7
2,5,8
1,3,7,8
2,3,7
Item sets
1
2
3
5
7
8
Support
3
4
5
3
4
3
Item sets
{1,3}
{1,7}
{2,3}
{2,5}
{3,7}
Support
3
3
3
3
4
Item sets
{1,3,7}
Support
3
Fig 2: An example of transaction data stream and find out maximum frequent item set.
Online mining of frequent item sets within a transaction-sensitive sliding window:
In this part we describe our algorithm that is a single pass mining algorithm called MFI-Trans SW
and its bit representation of items with the help of the example .In this we compared our
algorithm with the existing sliding window based mining techniques and improve the speed as
well as memory by dynamically maintenance in the current sliding window by the effective bit
sequence representation.
1) Bit-Sequence Representation of Items
Transactions Items Bit-
Sequence
T1<1,3,4,5>
T2<2,3,5,6,7>
T3<1,2,3,5,7>
1
2
3
4
5
6
7
8
101
011
111
100
011
010
101
010

5
101 &111=101
This is the bit
representation
for (1,3)
Fig: 3 Bit-Sequence Representations of Items.
The above figure 3 shows that the bit sequence representation. How we can represent our item set
in bit sequence representation. The computation of frequent item sets can be done efficiently by
representing all the items in the transactions of the data stream by bit sequence representation
[12]. If there are three transactions in the given window size then the length of the bit sequence
representation will be three. If there are four transactions then the length will be four. The first
bits represent the first transaction, second bit represent the second transaction and so on. If the
sliding window is containing three transaction and they contain different item set as shown in the
above fig. The concept of bit sequence representation are illustrated in the above figure 2 where
the bit sequence representation for items (1,3) is obtained by performing bit AND operator on bit
the sequence representation of item 1 and the sequence representation of item 3.The same
procedure is repeated for deriving the bit sequence representation of other item set of length 2
.Once the 2 item set are available then easily we can illustrate the three bit representation of item
three and so on. The bit sequence representation enables the computation of frequent item sets
easily because counting the number of 1’s in the bit sequence will give us the exact count of
number of transactions that contain the item set. This count can be compared with the support
threshold to find if the item set is frequent or not.
3. MFI- TransSW ALGORITHIM
3.1 Window Initialization Phase
This phase is activated when the number of transaction data stream is less than or equal to a user
defined sliding window size. In this phase every transaction is converted into bit sequence
representation.
3.2 Window Sliding Phase
ID ITEM SET Window (1, 2, 3, 4)
T1 1,3,4,7
T2 2,3,5,8
T3 1, 2,3,5,7
T4 2, 5, 8
T5 1,3,7,8
T6 2, 3, 7
Time line Window (1, 2, 3, 4)
Fig: 4 running example of window size 3.

6
Input: TDS (a transaction data stream),ms( user defined minimum support in the range of
[0,1,2],and ws (the user defined sliding window size)
Output: FI-Output (set of frequent item set)
Begin
/*the following algorithm is for sliding window how the transactions are going according
to GWS ( given window size) */
Item_count=0,t_count=0,GWS(given window
size),TDS Load
Tdsw=NULL; /*consist of number of
transactions */
For (i=0; i>=EOF TDS;i++) /*for each
transaction */
T_count++;
If (t_count>=GWS) then
W1=TDS (1) to TDS(GWS)
Same repetitions for W2
end
if (w2 > GWS) then
go to w3
end
/*the following algorithm is for candidate generation phase and find out the maximum
frequent item set */
do item = item of transaction
For (i=o; i<=EOF TDS;i++)
If (item ==TDS (i)) then
I_count =I_count+1;
end
if (I_count >=MS) then
CG1 = w2 (FI (k-1)/*candidate item generation step */
if (CG1>=MS) then
CG2=w3=FI (K) {CG2>=MS}
Item = w3;
Go to step 1;
While EOF TDS
end if
end for
Output = MFI (maximum frequent item set)
4. EXPERIMENTS
In this experiment we shows the result of proposed MFITSW (maximum frequent item set
transaction sensitive sliding window) algorithm .All the program are implemented using
Microsoft visual studio 2010(DOT NET FRAMEWORK 4.0) and performed on a 4 GHz Intel i3
PC machine with 512 MB memory running on windows 7.
From the description of the algorithm, it is clear that the mining MFI maximum frequent item set
over online is very efficient by maintaining the various item set in the various transactions. We
have to implemented our sliding window algorithm program in the DOT NET FRAMEWORK
4.0 MICROSOFT VISUAL STUDIO 2010 .We have to compare and testing frequent item set
mining over MFITSW maximum frequent item set transaction sensitive sliding window algorithm
implementation using with IBM Synthetic Data Generator proposed by Agrawal and Srikant.

7
Fig:2 shows the synthetic data stream, denoted by T5.I4.D1000K of size of 1 million transactions
has an average of 5 of items(T5) with average maximum frequent item set size of 4 items (I4).
Fig:-5 transactions data stream.
Fig:3 shows the significantly outperforms for memory consumption and CPU cost and take less
time to take out the maximum frequent item set of an online data stream. The sliding windows are
based upon the user requriments.If the user gives the value of GWS (given window size).
In this experiment first of download the synthetic data steams using IBM synthetic data generator
purposed by Agrawal and Srikant. After that converts the data stream into the CSV format and
run this CSV data stream into Microsoft visual studio framework 2010 and first of all design the
program in the design file drag and drops the label, text and button from the toolbox and make the
code or programs in the source file after that running the programs and get the output. Select the
CSV file of online transaction data streams and click on the submit button for find out the
maximum frequent item set as an output.
Fig:-6 output as maximum frequent item set.

8
Experiment 1
Fig: 7 number of transactions and processing time in ms.
Figure 7 shows that the graph between the processing time in mille second and the number of
incoming transaction. The new MFI-TransSW takes less time than the MFI-TransSW.
Experiment 2
In the experiment 2 we compare our algorithm on the basis of the memory (MB) and the number
of incoming sliding window in the generating frequent item set phase. This experiment shows that
purposed algorithm take little bit less memory then the existing one. We have performed my
algorithm in Dot Net framework. This algorithm increase the processing time as well as take less
time for generating
Fig: 8 incoming sliding window and memory uses infrequent item set generation phase.
5. CONCLUDING REMARKS
As we know that find out the maximum frequent item set of online transaction data stream is very
important concept .In this paper proposes a method for finding recently frequent item sets over a
data stream based on MFITSW mining frequent item set transaction sensitive sliding window
algorithm. In order to support various needs of data stream analysis, experiment shows that the
proposed not only attained the highly accuracy mining result but also run significant faster and
consume less memory then the existing algorithms for mining frequent item sets. In this paper the
interesting recent range of a data stream is defined by the size of a window. The proposed method
can be employed to monitor the recent change of embedded knowledge in a data stream. The
proposed algorithm gave a guarantee of the output quality and also a bound on the memory usage.
0
10
20
30
2000 4000 6000 8000
MFI-
TransSW
new MFI-
TransSW
30
35
40
45
New MFI-
Trans
MFI-
TransSW

9
REFERENCES
QuestDataMiningSyntheticDataGenerationCode.Available:http://www.almaden.ibm.com/cs/projects/iis/hd
b/Projects/datamining/datasets/syndata.html:-.
[1] B. Babcock, S. Babu, M. Datar, R. Motwani, and J.Widom. Models and Issues in Data Stream
Systems. In Proc. of the 2002 ACM Symposium on Principles of Database Systems (PODS 2002),
ACM Press, 2002.
[2] H.-F. Li, S.-Y. Lee/Expert Systems with Applications 36 (2009) 1466–1477.
[3] Han, J., Pei, J., Yin, Y., & Mao, R. (2004). Mining frequent patterns without candidate generation: A
frequent-pattern tree approach. Data Mining and Knowledge Discovery, 8(1), 53–87.
[4] Roberto Bayardo. Efficiently Mining Long Patterns from Databases. In ACM SIGMOD Conference,
1998.
[5] J. Chang and W. Lee. Finding Recent Frequent Item sets adaptively over Online Data Streams. In
Proc. of the 9th ACM SIGKDD International Conference & Data Mining (KDD-2003), 2003.
[6] V. Ganti, J. Gehrke, and R. Ramakrishnan. Mining Data Streams under Block Evolution. SIGKDD
Exploration, 3(2):1-10, Jan. 2002.
[7] C. Giannella, J. Han, J. Pei, X. Yan and P. S. Yu. Mining Frequent Patterns in Data Streams at
Multiple Time Granularities. In Proc. of the NSF Workshop on Next Generation Data Mining, 2002.
[8] G. S. Manku and R. Motwani. Approximate Frequency Counts Over Data Streams. In Proc. of the
28th VLDB conference, 2002.
[9] W.G. Teng, M.-S. Chen and P. S. Yu. A Regression Based Temporal Pattern Mining Scheme for Data
Streams.
[10] International Journal of Data Mining & Knowledge Management Process (IJDKP) Vol.2, No.6,
November 2012.
[11] J. Han, H. Cheng, D. Xin, & X. Yan. (2007) “Frequent pattern mining: current status and future
directions”, Data Mining and Knowledge Discovery, vol. 15(1), pp. 55–86.
[12] Hua-Fu Li, Chin-Chuan Ho, Man-Kwan Shan, and Suh-Yin Lee, (October 8-11, 2006) ” Efficient
Maintenance and Mining of Frequent Item sets over Online Data Streams with a Sliding Window”,
IEEE international Conference on Systems, Man, and Cybernetics Taipei, Taiwan.
[13] M. M. Gaber, A. Zaslavsky, and S. Krishnaswamy,(2005) "Resource aware Mining of data streams”,
Journal of universal computer science,vol II ,pp 1440-1453.
[14] K.FJea,C ,W Li,T P Chang, ”An efficient approximate approach to mining frequent item sets over
high speed transactional data streams “ ,Eighth International conference on intelligent system design
and applications, DOI 10.1109/ISDA2008.74 .
[15] Jia Ren, Ke LI , (July 2008) ” Online Data Stream Mining Of Recent Frequent Item sets Based On
Sliding Window Model “,Proceedings of the Seventh International Conference on Machine Learning
and Cybernetics, Kunming.
[16] J.H. Chang and W.S. Lee, (2003).Finding recent frequent item sets adaptively over online data
streams. In Proc. of the 9th ACM SIGKDD, pp. 487-492.
[17] D Lee, W Lee. (2005). “Finding Maximal Frequent Item sets over Online Data Streams Adaptively”,
Fifth Intl. Conference on Data Mining.
[18] Mozafari, H. Thakkar, and C. Zaniolo. Verifying and mining frequent patterns from large windows
over data streams. In Proc. Int. Conf. on Data Engineering, pages 179–188, Los Alamitos, CA, 2008.
IEEE Computer Society.
[19] Caixia Meng, (2009)”An efficient algorithm for mining frequent patterns over high speed data
streams”, World congress on software engineering.
[20] H.-F. Li, S.-Y. Lee, M.-K. Shan, An efficient algorithm for mining frequent item sets over the entire
history of data streams, in: Proc. International workshop on Knowledge Discovery in Data Streams,
2004.
[21] C.K.-S. Leung, Q.I. Khan, and DSTree: a tree structure for the mining of frequent sets from data
streams, in: Proc. ICDM, 2006, pp. 928–932. [23] C.K.-S. Leung, Q.I. Khan, Efficient mining of
constrained frequent patterns from streams, in: Proc. 10th International Database Engineering and
Applications Symposium, 2006.
[22] H.-F. Li, S.-Y. Lee, Mining frequent item sets over data streams using efficient window sliding
techniques, Expert Systems with Applications 36 (2009) 1466–1477.

10
[23] G.S. Manku, R. Motwani, Approximate frequency counts over data streams, in: Proc. VLDB, 2002,
pp. 346–357.
[24] B. Mozafari, H. Thakkar, C. Zaniolo, Verifying and mining frequent patterns from large windows
over data streams, in: Proc. ICDE, 2008 pp. 179–188.
[25] J. Li, D. Maier, K. Tuftel, V. Papadimos, P.A. Tucker, No pane, no gain: efficient evaluation of
sliding-window aggregates over data streams, SIGM ODRecord 34 (1) (2005) 39–44.
[26] C.-H. Lin, D.-Y. Chiu, Y.-H. Wu, A.L.P. Chen, Mining Frequent item sets from data streams with a
time-sensitive sliding window, in: Proc. SIAM International Conference on Data Mining, 2005.
[27] J.X. Yu, Z. Chong, H. Lu, Z. Zhang, A. Zhou, a False negative approach to mining frequent item sets
from high speed transactional data streams, Information Sciences 176 (14) (2006) 1986–2015.
[28] J.X. Yu, Z. Chong, H.Lu, A. Zhou, False Positive or false negative: mining frequent item sets from
high speed transactional data streams, in: Proc. VLDB, 2004, pp. 204–215.
[29] S.K. Tanbeer, C.F. Ahmed, B.-S. Jeong, Y.-K. Lee, CP-tree: A tree structure for single-pass frequent
pattern mining, in: T. Washio Et al. (Eds.), Proc. PAKDD, 2008, pp.1022–1027.
[30] Y.J. Tsay, J.Y. Chiang, An efficient cluster and decomposition algorithm for mining association rules,
Information Sciences 160(1–4) (2004) 161–171.
.
AUTHORS
Neeraj received B.E (Information Technology) from DAV College of Engineering and
Technology from Kanina (Haryana), INDIA. During 2006-2010 & pursuing M-tech
(Software Engineering) from ITM University Gurgaon (INDIA) during 2011-2013.

Mining Maximum Frequent Item Sets Over Data Streams Using Transaction Sliding Window Techniques

More Related Content

What's hot (6)

Viewers also liked (16)

Similar to Mining Maximum Frequent Item Sets Over Data Streams Using Transaction Sliding Window Techniques (20)

Recently uploaded (20)

Mining Maximum Frequent Item Sets Over Data Streams Using Transaction Sliding Window Techniques