Connected lighting system data

With the exponential increase of connected products and cheap digital data storage capabilities, applying data analysis and machine learning techniques on commercial datasets has been playing an important role to understand customer behavior. This understanding helps commercial decision-making in various aspects such as suitable time for product promoting or products placement in supermarkets. Nowadays, Intelligent lighting systems generate large amount of data that include product information, usage patterns and timestamp, etc. Such data can be used to solve several interesting business problems. In this study, three business questions are proposed based on business interests and data available, then answered by analyzing a sample dataset derived from real-world data of an connected lighting system. An ad-hoc data analysis framework is built for this aim. The framework is based on frequent pattern mining, which discovers frequent patterns in a given dataset. Three branches of frequent pattern mining: association rule mining, sequential pattern mining and frequent episodes mining are introduced to answer each question respectively. The generated patterns provide a clear understanding of the transactions happening in the data, which also provide a potential for decision making when using factual commercial and system data.