A near real-time GPS data analysis application. Luka receives location tracepoints from a smartphone and formulates a location-based timeline that is used to send actionable prompts.
What it does
Luka uses a novel algorithm to accurately group location data streams into activity stops despite a high data influx rate and spatial disparities. It is written in GoLang.
We have developed a novel algorithm that is able to cluster location streams in a heterogeneous dataset and produce results within milliseconds. The Luka API allows users to stream data for continuous analysis and subscribe to new location events.
Manual visualisation of the Luka analysis algorithm. Luka receives GPS tracepoints and interprets locations visited (in numerical order, not semantically), routes, time spent at places, your daily reach, and more.
Why we developed it
Location affects people's behaviour, so we use Luka to customise our Flex programme to your location trends. To do so, we analyse GPS tracepoints from your smartphone.
Existing methods such as DBSCAN and its variants do a good job in accounting for trajectory sequences with uncontrolled samples. However, this is not excellent for real-time analysis or applications. They do not solve the problems with GPS inaccuracies and misclassification of activities. In addition, the memory and CPU requirements to run these algorithms increases exponentially with the volume of data making it expensive to maintain.