System and method providing association rule aggregates

A relational database system, comprising:

• a processor;
• a data cache configured to store records of transaction data, wherein the records of transaction data represent multiple instances of transactions with respect to a plurality of items each represented by an item identifier and at least one metric value;
• a counting module stored in a non-transitory computer-readable medium including instructions that when executed cause the processor to perform:
• reading each of the records of transaction data to identify occurrences of at least one candidate item set, and
• counting a number-of-occurrences of the at least one candidate item set to generate a total count, wherein the at least one candidate item set is under consideration for being at least one frequent item set in the records of transaction data;
• an aggregation module stored in the non-transitory computer-readable medium including instructions that when executed cause the processor to perform, as the counting is proceeding through the records of transaction data, simultaneously generating of at least one aggregate metric value from corresponding records associated with the at least one candidate item set, wherein the generating is performed at least in part by summing metric values across the number-of-occurrences for each item represented in the at least one candidate item set;
• wherein simultaneously generating the at least one aggregate metric value includes accumulating, for at least one item represented in the at least one candidate item set, at least one of:
• a revenue value representing a revenue received for the at least one item in a transaction represented in the set of transaction data;
• a profit value representing a profit made on the at least one item in the transaction represented in the set of transaction data;
• or
• a quantity value representing a quantity of the at least one item purchased in the transaction represented in the set of transaction data;
• a level filtering module stored in the non-transitory computer-readable medium including instructions that when executed cause the processor to perform a filtering operation to:
• exclude the at least one candidate item set from being the at least one frequent item set at least in part by comparing the total count to a specified threshold value, or
• confirm that the at least one candidate item set is the at least one frequent item set at least in part by comparing the total count to the specified threshold value;
• an association rule module stored in the non-transitory computer-readable medium including instructions that when executed cause the processor to:
• for the at least one frequent item set determined, generate at least one candidate association rule that (i) associates at least two items from the at least one frequent item set in an if-then relationship and (ii) includes a confidence value that reflects a strength of association of the at least two items based at least in part on the total count;
• and
• a rule filtering module stored in the non-transitory computer-readable medium including instructions that when executed cause the processor to identify the at least one candidate association rule as a final association rule at least in part by comparing the confidence value of the at least one candidate association rule to a specified threshold probability level.

A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of a computing device, cause the computing device to at least:

• control a reading of a plurality of transaction records, via the one or more processors, wherein each transaction record of the plurality of transaction records represents a transaction for at least one item and includes transaction data that includes an item identifier and a metric value for the at least one item;
• control a counting of a number-of-occurrences of at least one candidate item set in the plurality of transaction records, via the one or more processors, to generate a total count for the at least one candidate item set, wherein the at least one candidate item set is under consideration for being at least one frequent item set in the plurality of transaction records;
• as the counting of the number-of-occurrences is proceeding through the plurality of transaction records, simultaneously control an accumulating of at least one aggregate metric value from corresponding transaction records associated with the at least one candidate item set, wherein the accumulating is performed at least in part by summing metric values across the number-of-occurrences for each item represented in the at least one candidate item set;
• wherein simultaneously controlling the accumulating of the at least one aggregate metric value includes accumulating, for at least one item represented in the at least one candidate item set, at least one of:
• a revenue value representing a revenue received for the at least one item in a transaction represented in the set of transaction data;
• a profit value representing a profit made on the at least one item in the transaction represented in the set of transaction data;
• or
• a quantity value representing a quantity of the at least one item purchased in the transaction represented in the set of transaction data:
• control a determination of whether the at least one candidate item set is the at least one frequent item set at least in part by comparing the total count to a specified threshold value via the one or more processors;
• wherein the determination of being the at least one frequent item set is performed simultaneously with the accumulating of the at least one aggregate metric;
• for the at least one frequent item set determined, generating, via at least the processor, at least one candidate association rule that (i) defines and associates at least two items from the at least one frequent item set in an if-then relationship, and (ii) includes a confidence value that reflects a strength of association of the at least two items based at least in part on the total count;
• and
• after the generating of the at least one candidate association rule, at least identify, via the one or more processors, the at least one candidate association rule as a final association rule at least in part by comparing the confidence value of the at least one candidate association rule to a specified threshold probability level.