Please enable JavaScript.
Coggle requires JavaScript to display documents.
MIND MAP A Research Done Related to Proposed Topic (Main Idea A.2 …
MIND MAP A
Research Done Related to Proposed Topic
Main Idea A.1
Disadvantage of the Apriori algorithm
The algorithm requires n+1 scans to scan the whole database, where n is the length of the longest itemset (Sudha 2017).
Apriori is only limited for a small database (Sahu et al. 2015).
Multiple number of times in scanning is needed to scan transactions in a database for calculating frequent item using the Apriori algorithm (Sahu et al. 2015).
Apriori only defines the presence and absence of an item Sahu et al. 2015).
As the size of the database increases, the time complexity of database scans also increases (Vijayalakshmi and Pethalakshmi 2013).
The algorithm wastes time scanning the whole database for frequent itemsets (Najadat et al. 2013). The algorithm checks the database for many sets from candidate items and scans it repeatedly in order to find other candidate itemsets.Apriori will work inefficiently when scanning large databases because its memory capacity is limited with a large number of transactions (Najadat et al. 2013).
Apriori works inefficiently in terms of memory requirement when large numbers of transactions are considered (Aggarwal and Sindhu 2015).
The performance of the algorithm in dense data is also shown to decline due to the large number of long patterns (Kaur 2014).
Main Idea A.2
Approaches and ways to improve the Apriori algorithm
Improved the Apriori algorithm by using the concept of “set size” which is the number of items per transaction and “set size frequency” which is the number of transactions that have at least “set size” items (Abaya 2012).
The transaction reduction is an approach which reduces the wasted time that depends only on the scanning of some transactions (Altameem and Ykhlef 2018).
Hash-based is an approach which shows the improvement in the execution of time and the utilization of space (Altameem and Ykhlef 2018).
Partitioning is an approach that is used to distribute the entire dataset into the smaller subset (Altameem and Ykhlef 2018)
Sampling is an approach that mines a random sampled subset from the entire dataset (Altameem and Ykhlef 2018).
A Matrix file using MATLAB tools was created. In this file, given transactions can be saved (Alwa and Patil 2013).
A modified Apriori algorithm which decreased the times of scanning the database. It also optimized the process that generates candidate itemsets and facilitates the counting support (Altameem and Ykhlef 2018).
This algorithm employs top-down approach to discover regular itemsets from largest regular itemset to smallest regular itemset (Dutt et al. 2014).
The algorithm was improved through the pruning of frequent itemsets where the itemsets which did not satisfy the given conditions were removed (Kaur 2014).
An improved Apriori algorithm was able to reduce the amount of the classic Apriori's execution time (Prajapati and Sheth 2015).
An algorithm which first creates a matrix based on the given dataset of problem was proposed. Afterwards, it calculates the support by performing various matrix operations like ANDing the rows for generating frequent candidate keys according to given minimum support (Mohan and Rajpoot 2016).
The algorithm was further improved by cutting down unnecessary transaction records (Singh et al. 2013).
An approach which reduces the time consumed of Apriori for candidate itemset generation by scanning all transactions to generate L1 which contains the items, their support count and Transaction ID where the items are found and then uses L1 later as a helper to generate L2, L3 ... Lk (Najadat et al. 2013).
The Bit Array Matrix, was shown to work efficiently because it discovered the patterns faster and gave good results in large databases (Vijayalakshmi and Pethalakshmi 2013).
The number of transactions to be scanned was reduced, thereby reducing the time consumed in transactions scanning for candidate itemsets. This improved algorithm reduced the time spent by the original algorithm by 67.38% (Najadat et al. 2013).
The improved Apriori, T_Apriori was able to reduce the time consumed in scanning transactions for the generation of candidate itemsets and the number of transactions to be scanned (Yuan 2017).
A hash algorithm based on the AprioriTid algorithm was shown to solve frequent itemsets quicker than the classic algorithm (Yu-gang et al. 2018).