my data consists of a point on a grid which moves around through time. so, you can look at this as a trail through a plane, or a timeseries of (x,y) coordinates. imagine this as a bug moving around on a table. is there a way i could measure the "entropy" of this system? is there an information entropy measure that would be appropriate?
On 16 lis, 16:09, Dingus Waternloo <timf...@gmail.com> wrote:
> my data consists of a point on a grid which moves around through time. > so, you can look at this as a trail through a plane, or a timeseries > of (x,y) coordinates. imagine this as a bug moving around on a table. > is there a way i could measure the "entropy" of this system? is there > an information entropy measure that would be appropriate?
> thanks in advance
Entropy is a statistical function, where temperature corresponds to standard deviation. In your case, it is possible to label different positions by different symbols, and to read their serie as a text. Information entropy is applicable. kunzmilan [Moderators note: corrected version]
> On 16 lis, 16:09, Dingus Waternloo <timf...@gmail.com> wrote:
> > my data consists of a point on a grid which moves around through time. > > so, you can look at this as a trail through a plane, or a timeseries > > of (x,y) coordinates. imagine this as a bug moving around on a table. > > is there a way i could measure the "entropy" of this system? is there > > an information entropy measure that would be appropriate?
> > thanks in advance
> Entropy is a statistical function, where temperature corresponds to > standard deviation. In your case, it is possible to label different > positions by different symbols, and to read their serie as a text. > Information entropy is applicable. > kunzmilan > [Moderators note: corrected version]
Thank you. Any suggestions for how this may be accomplished with the following data formats?
I understand how standard deviation could be used for temperature, but any ideas for which specific equations to use?
1) A list of Cartesian coordinates (2xN array), representing a path
On Nov 16, 9:09 am, Dingus Waternloo <timf...@gmail.com> wrote:
> my data consists of a point on a grid which moves around through time. > so, you can look at this as a trail through a plane, or a timeseries > of (x,y) coordinates. imagine this as a bug moving around on a table.. > is there a way i could measure the "entropy" of this system? is there > an information entropy measure that would be appropriate?
> thanks in advance
If the movement of the bug is discrete, in the sense that there are N possible vectors describing the movement, each direction can be coded with a different letter in an alphabet of N characters. Each path is then a word made up of a finite sequence of letters from this alphabet. The entropy of such words is straightforward to compute. One could make the case that this entropy corresponds to the entropy of the bug's motion.
|
