Trying to create a OD matrix shapefile with data from centroids using Python

Joaquin_Osorio · ‎09-14-2021

Good morning.

I'm trying to create a OD matrix shapefile using Python. I can't use Network Analyst since I am using about 3100 Origin points and Destiny points. What I want is just the line shapefile, I don't care about distance or time travels.

I have used a script I found in this thread: https://community.esri.com/t5/data-management-questions/is-there-a-way-to-connect-all-points-in-a/td...

While it works and I have my OD matrix, it has no data from the centroids, and then preparing the data for the analysis I'm working I find the problem that spatial join with the centroids to enrich data generates two flows from point 1 to point 9 instead a flow from 1-9 and the other one to 9-1.

What I thought is to directly enrich the data in the step that creates the lines, so I can have a column with the name of the origin centroid and another one with the name of the destination centroid.

In my case, the fields I want to add to the OD matrix are called 'ID_GRUPO','SUM_POB_AS','NOMBRE_CEL'. I tried this script but the result I got is again a OD matrix with just a field called ID which has always a value 0.

fc = 'celdas_centroids' # feature class

sr = arcpy.Describe(fc).spatialReference # spatial reference

lines = [] # output line list

geoms = [i[0] for i in arcpy.da.SearchCursor(fc,['SHAPE@','ID_GRUPO','SUM_POB_AS','NOMBRE_CEL'],spatial_reference=sr)] # point geometries

for geom1 in geoms: # loop through points

for geom2 in geoms: # compare to points

lines.append(arcpy.Polyline(arcpy.Array([[geom1.centroid,geom2.centroid]]),sr)) # create line

arcpy.CopyFeatures_management(lines,r'F:\matriz_celdas_od.shp') # output to feature class

Thank you very much for the help.

DavidPike · ‎09-15-2021

Can you elaborate on this as I don't quite understand.

I find the problem that spatial join with the centroids to enrich data generates two flows from point 1 to point 9 instead a flow from 1-9 and the other one to 9-1.

Joaquin_Osorio · ‎09-15-2021

Yes. I've tried to create a OD matrix using the Python script from the other thread I posted using a shapefile with +3200 centroids. As I result I get a shapefile with +10 million lines with only one field called "ID" with value 0. What I want is that every line has incorporated some fields from the centroids shapefile.

A solution I thought was in doing a spatial join one to many of the empty created line shapefile with my centroids shapefile. I get as a result a shapefile with +20 million lines, in which every line has the data from a centroid. There is also a field created when I do the spatial join called FID_ID, in which two lines have the same FID_ID value (making it a OD pair). Then I can summarize data to create a table with a single FID_ID value, the ID of the origin point and the ID of the destination point.

While this gives a very close result to what I want, there is only one improtant problem: the order the lines ID have in the spatial join shapefile and the order the summarized table is created.

To put an example, I have this in the spatial join shapefile:

FID 14 Origin centroid: ID 3

FID 14 Destiny centroid: ID 5

FID 15: Origin centroid ID 4

FID 15: Destiny centroid ID 5

FID 16: Origin centroid ID 5

FID 16: Destiny centroid ID 6

FID 300 Origin centroid: ID 3

FID 300 Destiny centroid: ID 5

FID 301: Origin centroid ID 4

FID 301: Destiny centroid ID 5

FID 302: Origin centroid ID 5

FID 302: Destiny centroid ID 6

I want FID 14 being a OD flow 3-5, FID 15 being 4-5, and FID 16 being 6-5, and then FID 300 being 5-3, FID 301 being 5-4, and FID 302 being 5-6.

But due to the order in the table and in the data summarization I'm forced to have two flows 3-5, 4-5, and 5-6. I don't know how to be able to get that.

So I thought in trying to create directly the OD matrix from the script, since the script already assumes an origin centroid, and a destination centroid, while the spatial join just assumes two points in ID order.

I've slightly edited the code to this: