Calculate the arc of point data

[ChantelSteele]

Hello,

I have a point dataset that contains hundreds of thousands of points. I am wanting to combine points within approx 1km radius (using a like feature) and then calculate the arc/angle of those combined points. The points are generally clustered in a circular pattern, but rarely 360 degrees, so I am wanting to  calculate the angle of these clustered points to then eventually run statistics on their angles.

Apologies if this is vague - generally speaking my analysis skills are more limited and previously have used ArcGIS more in cartography than analysis so any help would be much appreciated.

Cheers

[DanPatterson_Retired]

Do you have an image to demonstrate your patterns?  Have you a predefined method to identify point clusters? There are several options to delineate the clusters, but once they are identified, are you wanting to determine the angles formed by their hull? or the angles, to the hull, from some type of centroidal point within the points?  Any visuals or details would help

[ChantelSteele]

Hi Dan,

Thank you for your assistance. Yes, I have (see below). You can see how the points are clustered in a circular formation. I'd like to be able to group these points, find the centroid and calculate the angle of the arc as well as the diameter (ie distance between the two furthest points)... These points are at approx. half hour intervals, per person. Each person may have more than one cluster of points and I am not interested in any points outside of these clusters so therefore, I cannot group by person. I hope this makes sense and would be grateful for your help.

[DanPatterson_Retired]

Long time ... no hear...

So you are interested in results like this?

Linear                                               

        ID,           Xl,           Yl,           Rl,         

         1, 341643.864709, 5024353.680128,   924.664323

Nonlinear

Xn,           Yn,           Rn

341694.871606, 5024305.018460,   992.982376

Where X,Y,R are the respective centers and radii of linear and Nonlinear circular regressions.

That look something like this?

I haven't finished elliptical regression yet, so be forewarned that some of the patterns look like ellipses would fit better than circles, however, a circle is closer to an ellipse, and is closer still than properties that can be derived from either the convex or concave hulls of the data (I have tested...so don't go there).

It can be done.  I wrote a toolbox that performs linear and non-linear circular regression for feature data.  Yours happens to be point data which is nice, however, it would be nice if the point clouds also be separated by groupings...I can do this manually by selecting what looks like a circular pattern and then returning the statistics and a polyline.

I am in the process of translating my 'old stuff' to new stuff and I am always looking for testing files, so if you want to share, I could have a run through your data...It might spur me on to look at it earlier than anticipated.  So if you want to contact me offline to send some sample data, check my email on my profile...it will go to my University email and I can deal with it from there.

PS... currently working in 10.3, but I can also test on 10.2.2 ... although there are no major issues between the two.