Specifically, this could be achieved by pre-computing which data points are within height_limit of each other and then identifying clusters in those rather than stepping through the data one at a time (which requires re-checking the same data point comparisons multiple times)
Specifically, this could be achieved by pre-computing which data points are within height_limit of each other and then identifying clusters in those rather than stepping through the data one at a time (which requires re-checking the same data point comparisons multiple times)