Description
Based on the topic of your project, find the initial references: Go to Google Scholar and search based on your topic: Provide at least 5 references.Explain shortly for each reference why you chose it. Describe the overall goal of your project: What are you trying to show?Which tools and what data do you plan to use?
Unformatted Attachment Preview
Project 5: Faster rule extractions in market basket analysis
Motivation:
The market basket analysis is focused on finding patterns in transactional databases. It seeks to
find the association rules among the fields in the records. Usually, the disadvantage of market
basket analysis is that the number of association rules in practice is too large, so the rule finding
is usually computationally intensive and time-consuming.
Paper purpose:
Evaluate for correctness and efficiency 2 procedures or algorithms that find the association rules,
so the algorithm is practical in terms of computational time and space cost.
Good methodologies to find relationships:
Compare 2 basic approaches: 1st build a spanning tree model base on the transaction
correlations (see Valle, Mauricio A., Gonzalo A. Ruz, and Rodrigo Morrás paper), 2nd build a graph
where vertices are binary encodings of each record, and the directed weighted edges based on
the Hamming distances between the binary encodings. To evaluate the 2 approaches, generate
2^16=65536 random records, each with 8, 1 or 0, fields. Each vertex is a unique binary record
and there are 256 vertices. E.g., Hamming distance of -1 would indicate an association rule in
which 1 item is predicted. To speed up the procedure use binary operations in your code.
Possible additional outcomes:
Find out the greatest size of the problem in the market basket analysis so the rules can be found
reasonably fast.
Starting help resources:
Market basket analysis using minimum spanning trees. by Valle, Mauricio A., Gonzalo A. Ruz, and
Rodrigo Morrás, Network Intelligence Meets User Centered Social Media Networks 4 (2018)
Bonanno, Caldarelli, G., Lillo, F., & Mantegna, and R. N. (2003). Topology of correlation based
minimal spanning trees in real and model markets.
Keywords:
Market basket analysis, Minimum spanning tree, Network of products, Association rules.
Purchase answer to see full
attachment
