Extracting Gradual Periodic Patterns from Transactional Databases: Application to Felonies and Misdemeanors Data

In recent years, discovering interesting periodic frequent patterns from transactional databases has become an attractive task for many researchers. It consists in finding patterns that are not only frequent in a database but also periodic, i.e., their successive occurrences are sufficiently close to each others. In this paper we propose a new methodology that analyses the combination of the results of two algorithms for mining periodic patterns in order to distinguish different levels of pattern’s periodicity and hence define a kind of gradual periodicity. This idea is then applied to a database of felonies and misdemeanours that occurred in the wilaya of Bordj Bou Arreridj in Algeria, within the years 2020-2021. The obtained results allow one to associate to each area of the wilaya its dangerousness level. This may be very helpful for police authorities in taking useful decisions.