patternsfound

Patternsfound is a metric used in pattern discovery and data mining to denote the total number of patterns that satisfy predefined criteria within a dataset or search space. It is a result of the algorithm and the specification of what counts as a pattern, including thresholds such as minimum support, minimum sequence length, or subgraph size. Consequently, patternsfound reflects both the data properties and the user-defined constraints rather than a fixed attribute of the data itself.

In different subfields, patternsfound is interpreted according to the type of patterns being mined. In frequent itemset mining, it equals the number of itemsets whose frequency meets the minimum support. In sequential pattern mining, it counts discovered sequences that satisfy support and additional constraints like gaps or length. In graph mining, it counts subgraphs that meet size and occurrence criteria. The term serves as a common yardstick for comparing results across algorithms, datasets, and parameter settings.

Interpretation and use: A high patternsfound value can indicate a rich or highly constrained search space, suggesting meaningful structure or, alternatively, the risk of combinatorial explosion if thresholds are too loose. Low values may reflect data sparsity or strict criteria. Analysts often report patternsfound together with runtime, memory usage, and pruning strategies to provide a comprehensive view of an algorithm’s performance.

Limitations: The count is sensitive to parameter choices and may not reflect pattern quality. Therefore, it is commonly complemented with measures of statistical significance, redundancy reduction, or pattern interestingness to provide a more robust assessment.

See also: Frequent itemset mining, Sequential pattern mining, Subgraph mining, Association rule mining, Apriori, FP-growth.