KatjaDellaLibera/ StickyDBSCAN

Determine fission-fusion dynamics from position data

Contributed by: Katja Della Libera

Paclet publishing the "sticky" DBSCAN-modification used in the animal behaviour paper "Rate of fission-fusion events in domestic sheep is linked to time of day but not resource distribution". The first function "StickyDBSCAN" allows a user to assign groups based on GPS (or other coordinate) data for a number of individuals and time steps. The second function "FindFissionFusionGraph" takes the output of DBSCAN and turns it into a graph that associates groups with overlapping membership if they exist in consecutive time steps.

Installation Instructions

To install this paclet in your Wolfram Language environment, evaluate this code:
PacletInstall["KatjaDellaLibera/StickyDBSCAN"]

Examples

Basic Examples

Sample 5 Brownian Motion paths. We will use this as mock GPS data for 5 individuals:

In[1]:=

Visualize the Paths:

In[2]:=

Out[2]=

Make a list of all the pairs:

In[3]:=

Out[3]=

For each of these pairs find the distance at every time step:

In[4]:=

allBrownianPairData = Map[Table[
EuclideanDistance[samples[[#[[1]], t]], samples[[#[[2]], t]]], {t, 1, Length[samples[[1]]]}] &, allBrownianPairs, {2}];

Use the stickyDBSCAN function to find the groups at each time step. Let's try an inner radius of 0.3 and an outer radius of 0.5. the tmin is 1 to start from the beginning and the tmax is the number of fake GPS data points:

In[5]:=