PRODUCT BUNDLING APPLICATION TO SUPPORT IMPLEMENTATION OF ASSOCIATION RULE MINING USING FP-GROWTH

Online media has been proven to gave good impact especially in the business field, for opening and widen access to people and society, and so giving more opportunity to any business broader and easy access through electronic commerce (e-commerce). This is the basic idea for the need to improve business reach for a decorative plants store in Tangerang Area. The association transaction pattern as the outcome of data mining process by implementing the Fp-Growth Algorithm that is injected to a mobile application with the purpose to get the buyer’s shopping pattern, which is also packed as the products bundling promotion media and finally open the opportunity to broaden reach for business and information. Experiments carried out using different minimum support and confidence values, which are (1% with 10%, 2% with 20%, 3% with 25%, and 4% with 30%) is giving results in the form of seven association rules obtained from experiments with a minimum support 4% and minimum confidence 30%). The resulting association rules show that buyers (customers / non-customers) tend to buy plants and their growing media.


INTRODUCTION
Every company has data stored in its database. The transaction data is increasing day by day. With the increasing amount of data in the company, the role of the analyst to analyze data manually needs to be replaced with computerbased applications. So that the analysis process can be carried out precisely and accurately. This study applies the FP-Growth algorithm in an application that can find out consumer buying patterns at each different branch with different characteristics. From the resulting pattern, information will be obtained. Association analysis is also known as one of the data mining techniques which is the basis of various other data mining techniques. In particular, one of the stages of association analysis called high frequency pattern analysis (frequent pattern mining) has attracted the attention of many researchers to produce efficient algorithms. The importance of an associative rule can be determined by two parameters, support (supporting value), namely the percentage of the combination of these items in the database and confidence [5].
Ecommerce E-commerce is a term that is often used or heard today that is related to the internet, where no one clearly knows the meaning of e-commerce. Electronic commerce or also known as e-commerce, is how people use communication networks and computers for business processes to be done. A popular outlook for e-commerce is the use of the internet and a computer to buy and sell products using web browser [6].
The benefits of using E-commerce in a company as a transaction system are: a. Can increase market exposure (market share) b. Reducing operating costs (operating cost) c. Extending reach (global reach) d. Increase customer loyalty e. Improve supply management A brief overview of the advantages of e-commerce is as follows: a. For consumers: lower prices, shopping in one place. b. For managers: efficiency, error-free, and on time.

Data mining
Data mining is an automatic search process for useful information in large data storage areas. Other terms that are often used include knowledge discovery (mining) in databases (KDD), knowledge extraction, data / pattern analysis, data archeology, data dredging, information harvesting, and business intelligence. Data mining techniques are used to examine large databases as a way to find new and useful patterns. Not all information search task are stated as data mining. For example, searching for individual records using a database management system or search for a specific web pages through queries to all search engines is an information search task that is closely related to information retrieval. Data mining techniques can be used to improve capabilities of the information retrieval systems [4]. The definition of association rule is a convention that has an interrelated relationship between attributes used to obtain associations between data. There are several terms of association, that is the antecedent representing "if" and consequent representing "then". In the association analysis, these two things are groups of items that are not related together [4]. Frequent Pattern Growth (FP-Growth) is an alternative algorithm that can be used to determine the most frequently occurring data set (frequent itemset) in a data set [2]. The FP-Growth algorithm is a development of the Apriori algorithm. So that the shortcomings of the Apriori algorithm are corrected by the FP-Growth algorithm [2].
FP-Growth uses the concept of tree construction in searching for frequent itemsets. This is what causes the FP-Growth algorithm to be faster than the Apriori algorithm. The characteristic of the FP-Growth algorithm is that the data structure used is a tree called FP-Tree. By using FP-Tree, FP-Growth algorithm can directly extract frequent itemset from FP-Tree.

Jurnal Teknologi Informasi dan Pendidikan
Volume 14, No. 1, Maret 2021 tip.ppj.unp.ac.id FP-tree is a compressed data storage structure. FP-tree is built by mapping each transaction data into each particular path in the FPtree. Because in each mapped transaction, there may be transactions that have the same item, so the paths allow for overlapping. The more transaction data that has the same item, the more effective the compression process with the FP-tree data structure will be.
Mining of frequent itemsets using the FP-Growth algorithm will be carried out by generating a tree data structure (FP-Tree). The FP-Growth method can be divided into 3 main stages, namely as follows [4]: a. Generation Phase conditional pattern base Conditional Pattern Base is a subdatabase that contains prefix path (prefix path) and suffix pattern (suffix pattern). The conditional pattern base generation is obtained through the previously built FP-tree. b. FP-Tree conditional generation stage At this stage, the support count of each item in each conditional pattern base is summed, then each item that has a greater support count equal to the minimum support count will be generated with the conditional FP-tree c. Frequent itemset search stage If the conditional FP-tree is a single path, then the frequent itemset is obtained by combining items for each conditional FP-tree. If it is not a single path, then the FP-growth generation is done recursively.

Android
Android is an operating system for mobile devices that includes an operating system, middleware and applications. Android provides an open platform for developers to create their applications. Android as "the first complete, open, and free mobile platform" Complete (Complete Platform): android is a safe operating system and provides many tools in building software and allows for open application development opportunities (Open Source Platform) : developers can freely to develop Free applications (Free Platform): android is a platform / application that is free to develop. There is no license or royalty fee to be developed on the android platform. The minimum support count of 126.12 rounded down to 126.

Formation of a Header
Formation of frequent itemset a. Minimum support 3% and minimum confidence 25% The experiment was carried out 16 times with several minimum support values and