AUTHOR=Upadhya K. Jyothi , Lobo Ronan , Chhabra Mini Shail , Paleja Aman , Rao B. Dinesh , M. Geetha , Sisodia Prachi , Reddy Bolusani Akshita 

TITLE=Sliding window based rare partial periodic pattern mining algorithms over temporal data streams

JOURNAL=Frontiers in Big Data

VOLUME=Volume 8 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2025.1600267

DOI=10.3389/fdata.2025.1600267

ISSN=2624-909X

ABSTRACT=Periodic pattern mining, a branch of data mining, is expanding to provide insight into the occurrence behavior of large volumes of data. Recently, a variety of industries, including fraud detection, telecommunications, retail marketing, research, and medical have found applications for rare association rule mining, which uncovers unusual or unexpected combinations. A limited amount of literature demonstrated how periodicity is essential in mining low-support rare patterns. In addition, attention must be placed on temporal datasets that analyze crucial information about the timing of pattern occurrences and stream datasets to manage high-speed streaming data. Several algorithms have been developed that effectively track the cyclic behavior of patterns and identify the patterns that display complete or partial periodic behavior in temporal datasets. Numerous frameworks have been created to examine the periodic behavior of streaming data. Nevertheless, such a method that focuses on the temporal information in the data stream and extracts rare partial periodic patterns has yet to be proposed. With a focus on identifying rare partial periodic patterns from temporal data streams, this paper proposes two novel sliding window-based single scan approaches called R3PStreamSW-Growth and R3PStreamSW-BitVectorMiner. The findings showed that when a dense dataset Accidents is considered, for different threshold variations R3P-StreamSWBitVectorMiner outperformed R3PStreamSW-Growth by about 93%. Similarly, when the sparse dataset T10I4D100K is taken into account, R3P-StreamSWBitVectorMiner exhibits a 90% boost in performance. This demonstrates that on a range of synthetic, real-world, sparse, and dense datasets for different thresholds, R3P-StreamSWBitVectorMiner is significantly faster than R3PStreamSW-Growth.