We use our proprietary pattern mining software to find existing patterns in parsed data. We look for association rules between customized consumer variables. There is significant variability in human activity that can be observed from online data through a repeating structure over long term use on both social media and online shopping portals. As such, we are able to dive into a discovery of rich routine information and apply it to a number of applications to give a comprehensive trend analysis output.
Our pattern mining algorithm incorporates temporal information by mining multiple user stated data streams from various digital platforms at different granularities. It constructs a prefix-tree of patterns, where each stream specifies support for a pattern over time amongst your targeted consumer demographic.