I want to know more about visualization and density estimations.
Basically, I have a large sample of location data and every location object has a duration in milliseconds.
I want to solve the problem: "How long a person is staying at that point" using heatmap visualization.
I have read about Kernel Density Estimation where data points are being grouped in bin-grids or kernels.
Thus, I would like to ask 2 questions:
- Should I calculate Standard Deviation (SD) of duration for each point or as a whole (i.e. with relation to other points)?
- How do I "transform" or "combine" SD of each point to a KDE?
Overall, I'm planning to do additive blending using Gaussian kernels as stated in this reference.
Thank you all for your help!