A recurring event is an event that occurs at regular time intervals. The detection of such events can become a difficult task, especially when it comes to covering all possible recurring scenarios. Knowing the recurrent events and their periods allows us to predict their behavior in the future and therefore minimize the surprises and better planning our actions.
For companies like banks, insurances, corporations, etc. knowing the recurring behavior in customers’ data can help understand the needs of each branch to have a better plan to support their clients (i.e. planning their loans), and/or stop them from illegal actions (i.e. money laundering). Plus, it is a way for those companies to target categories of customers for their proper offers and advertisements.
In this project, we look for recurring transactions, that is to say, a collection of transactions for an account or a card that happens regularly with almost the same amount. Several techniques have been developed to store, read, visualize and analyze the data, with detecting such recurring behavior, especially for big data. Techniques from rule-based to machine learning that have been recently developed are explained and their advantages and lacks are mentioned in this project. In this project, we aim to investigate different methods for data visualization, analysis, and recurring detection.