DMA/L/z2

< DMA | L

In class

Unzip the file transactionsApriori.zip and read its content in MATLAB (use 'load' command).
Get familiar with the data format according to which shopping transactions are kept.
Get familiar with the java.util.Hashmap type (from MATLAB's command line test methods: put, get, containsKey, keySet, etc.)
Implement as a script the first part of the Apriori algorithm: finding frequent sets (for given minSupp parameter). It is convenient to return the frequent sets as a cell array of hash maps.
Implement as a script the second part of the Apriori algorithm: generation of association rules (using as input the result from the previous script).

Carry out 9 experiments for all (minSupp, minConf) pairs, where minSupp = 0.2, 0.15, 0.1; and minConf = 0.6, 0.8, 0.9. Apart from the list of rules, the result of each experiment should also be a plot of rules (command: scatter) marked as points in the supp x conf coordinate system.

Create a script taking the list of rules as an input and searching for equivalances among them (implication working in both directions), e.g.: if a -> b and also b -> a, then a <-> b. For each equivalance being displayed, write out the condifences for both directions of implication. Do the search for equivalances for all 9 experiments.

To speed up the checking process of your work at next lab (to avoid rerunning of 9 experiments), save the results of your homework in MATLAB into .mat files (using 'save' command). Note that the plots can also be assigned to variables and saved.