HW07 (27.03) - Association rules, ...
Use the data about transactions in a supermarket. Run FIM and Association rule generation algorithms to identify interesting itemsets and rules.
DATA file - Attach:supermarket.txt
1. Report which tools you decided to use, how you used them, what were the first results. Also report the running times for the tools chosen.
2. Report overall 5 different high-support, high-confidence, high-lift rules; provide the respective contingency tables and scores.
3. Discuss whether some other scores studied last week or in the lecture slides would help identify "more interesting" and different rules?
4. Given the ability to discover frequent itemsets and association rules, propose a strategy to use these tools to study different customer segments, shops, shopping times, or specific products.
5. Select some relatively high-support high-confidence rule (A->B) and based on that example describe the conditional probabilities P(A|B) and P(B|A), as well as the Bayes rule.
6. (Bonus 2p) Run Krimp on same data, provide commands and describe your findings and compare to FIM+Association rules. (Krimp documentation)