International Journal of Computer Applications (0975 – 8887) Volume 165 – No.12, May 2017 14 Analysis of Sequential Mining Algorithms Surbhi Chandhok Bachelors of technology Computer Science Galgotias College of Engineering & Technology Uttar Pradesh, India Romil Anand Bachelors of Technology Computer Science Galgotias College of Engineering & Technology Uttar Pradesh, India Soumay Gupta Bachelors of Technology Computer Science Galgotias college of Engineering &Technology Uttar Pradesh, India Aatif Jamshed Masters of Technology Computer Science Galgotias college of Engineering &Technology Uttar Pradesh, India ABSTRACT This paper essentially analyses the sequential pattern of mining algorithms. The discovery of Association relationship seeks more attention in data mining due to the constantly increasing amount of data stored in the real application system. Mining for association rules has its usage in several areas of business such as the process of decision making and the development of customized marketing programs & strategies. Therefore, the primary objective of data mining is to transform “data into knowledge”. As a result of which, mining association rules from enormous databases has been a significant topic in recent research for knowledge discovery in databases. It is known that database can be both dynamic and static. Static databases are the ones that do not change or alter with the passage of time. On the other hand, in dynamic databases, various new transactions append as time passes by. This might result in the production of some new itemsets while it is possible that certain frequent itemsets might as well become invalid. Therefore, in dynamic databases, the maintenance of large itemsets can be extremely expensive, in case rerun of previous mining algorithms on updated database is applied as it repeats a major portion of work done during previous computations. Apart from this, there is also lack of space for the storage of all the data and its processing. Therefore, it is recommended that instead of finding enormous itemsets again, certain heuristics be used for mining of dynamic databases. It brings forth the study of sequential pattern- mining algorithms, classified into five varied classes. 1. on the basis of Apriori-based algorithm. 2. on the basis of FP-Growth Algorithm. 3. on the bassis of Fast Algorithm. 4. on Partition Based Algorithm. 5. on the basis of Fast Update algorithm. Keywords Sequential Pattern, Data Mining, Pattern analysis, 1. INTRODUCTION It can be said that the topic of Sequential Pattern mining is primarily concerned with looking for statistically approved patterns between different data examples where the valuesare delivered in a sequence. Usually, it is presumed that the values are distinct and therefore, we can say that time series mining is also closely related and similar, but it's usually considered a separate activity. Sequential pattern mining is also known as a special case of structured data mining. 2. CATEGORIS OF PATTERN MINING ALGORITHM SEQUENTIALLY Algorithms for sequential pattern mining can be classified into the following classes:- 1. Apriori-likealgorithms 2. BFS Breadth First Search)-based algorithms 3. DFS (Depth First Search)-basedalgorithms 4. closed sequential pattern based algorithms 5. incremental-based algorithms 2.1 Static Datamining Static Data Mining can be defined as one that uses static database for mining. There can be a wide range of static data mining algorithms such as Fp- Tree, Partition based algorithm, Apriori, Fast algorithmetc. 2.1.1. Apriori Algorithm The most widely accepted static data mining algorithm- Apriori is often described as a “fast algorithm used for mining association rules”. It is enforced by marketbasket data. Also, it effectively produces large itemsets along with candidate itemsets through the process of repeatedly scanning the database. Apriori algorithm is the one based on candidate set generation accompanied with the test method. The unresolved issue that often appears during the process of mining frequent relations is, repeated scanning of original database, large number of candidate generation along with workload of support counting of the candidates. Hence, there is a need to begin reduction of passes of transaction database scans, in order to lessen the number of candidates and to help with support counting of candidates. Apriori algorithm isn't efficient enough, regardless of it being the building ground for several efficient algorithms. Suppose there is a transaction database which includes customer sequences. This database is composed majorly by three attributes: 1. customer-id 2. transaction time 3. purchased-item