International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017
p-ISSN: 2395-0072
www.irjet.net
PROFITABLE ITEMSET MINING USING WEIGHTS T.Lakshmi Surekha1, Ch.Srilekha2, G.Madhuri3, Ch.Sujitha4, G.Kusumanjali5 1Assistant
Professor, Department of IT, VR Siddhartha Engineering College, Andhra Pradesh, India. Department of IT, VR Siddhartha Engineering College, Andhra Pradesh, India. 3Student, Department of IT, VR Siddhartha Engineering College, Andhra Pradesh, India. 4Student, Department of IT, VR Siddhartha Engineering College, Andhra Pradesh, India. 5Student, Department of IT, VR Siddhartha Engineering College, Andhra Pradesh, India. ---------------------------------------------------------------------***--------------------------------------------------------------------2Student,
Abstract - In recent years, a number of association rule
mining algorithms like Apriori and FP- Growth were developed. But they are purely binary in nature. They do not consider quantity and profit (profit per unit). In these algorithms, two important measures viz., support count and confidence were used to generate the frequent item sets and their corresponding association rules. But in reality, these two measures are not sufficient for decision making in terms of profitability. In this a weighted frame work has been discussed by taking into account the profit (intensity of the item) and the quantity of each item in each transaction of the given dataset. Apriori and FP Growth algorithms are the best algorithms to generate frequent item sets, but they do not consider the profit as well as the quantity of items in the transactions of the database. Here we propose to new algorithms Profitable Apriori and Profitable FP Growth in our project which eliminate the disadvantages of traditional association rule mining algorithms and they also consider quantity and profit per unit. In this by incorporating the profit per unit and quantity measures we generate the most Profitable Itemsets and we compare the results obtained by Profitable Apriori and Profitable FP-Growth.
Key Words: Profit, Quantity, Profitable Item sets, Profitable Apriori, Profitable FP Growth.
1. INTRODUCTION Mining frequent patterns or Itemsets is an important issue in the field of data mining due to its wide applications. Traditional Itemset mining is, however, done based on parameters like support and confidence. The most widely used algorithms to obtain frequent Itemsets are Apriori and Frequent pattern growth. They are binary in nature. It means that they only consider whether the product is sold or not. If the product is sold, then it is considered true and else false. And these algorithms produce frequent itemsets, which only consider the occurrence of items but do not reflect any other factors, such as price or profit. Profitable Itemset Mining has recently been proposed, in which transactions are attached with weighted values according to some criteria. It is important because if support and confidence are only the parameters assumed, we may miss some of the profitable patterns.. However, the actual significance of an Itemset cannot be easily recognized if we do not consider some of the aspects like quantity and profit per each item.. The problem of Profitable Itemset mining is to find the complete © 2017, IRJET
|
Impact Factor value: 5.181
|
set of Itemsets satisfying a minimum profit constraint in the database. When we are calculating the Profitable Itemsets we can consider minimum weight as constraint and we can ignore the support as our goal is to find the Profitable patterns. In the real world, there several applications where specific patterns and items have more importance or priority than the other patterns. Profitable Itemset mining has been suggested to find Profitable patterns by considering the profits as well as quantity of Items. The concept of Profitable Itemset mining is attractive in that profitable patterns are discovered. We can use the term, Profitable Itemset to represent a set of profitable items. The Itemsets we get are frequent profitable itemsets as well as infrequent profitable itemsets.
1.1 BASIC CONCEPTS Itemset mining helps us to find the frequent patterns or itemsets . The two most widely used algorithms are Apriori and FP Growth. These two algorithms are binary in nature. They concerned about whether the product is sold or not. The measures considered by these algorithms are support and confidence. But in reality they are not sufficient for decision making in the large organizations. So In this framework we consider two measures named Quantity and Profit. By using both the parameters we calculate Weight. Consider the following two transactions: T1: {20 Buns, 5 Chocolates} T2: {1 Bun, 1 Chocolate} In the support-confidence frame work the above two transactions are considered to be the same, since the quantity of an item is not taken into account. But in reality, it is quite clear that the transaction T1 gives more profit than the transaction T2. Thus to make efficient marketing we take in to account the quantity of each item in each transaction. In addition we also consider the intensity of each item, which is represented using profit per item p. Consider the following two transactions: T3: {10 Buns, 1 Chocolate} T4: {2 Buns, 3 Chocolates} In reality the quantity sold in transaction T3 is greater than transaction T4, but the amount of profit gained by selling a chocolate (Say Dairy milk) is 10 times that of a Bun. So, the profit is also given priority represented by p. “p” may represent the retail price / profit per unit of an item. ISO 9001:2008 Certified Journal
|
Page 2604