I'm trying to spatially join 92,000+ points to see which census tract polygon it falls into (so I can get the GEOID for each point). It seems that either 1). ArcMap freaks out when the point is too close the the border of a tract and assigns it a random GEOID (possibly the correct GEOID, possibly the GEOID that is right next to it or 2). some points that are in the middle of the polygon are assigned the incorrect GEOID. When I performed a smaller batch it seemed to give the correct data. Does anyone have any thoughts on the number that ArcMap can handle? Thanks!
If the results are incorrect, the number is not relevant.
Are the data in a projected coordinate system?
Are you using the Spatial Join—Help | ArcGIS for Desktop and if so are you using intersect?
Have you just tried an Intersect?
Anything else that might account for the errors would be useful (ie are they in a particular area etc)
Thanks for your response, Dan! Yes, it's projected in USA_Contiguous_Albers_Equal_Area_Conic. I was just going about it by right clicking on the layer, then hitting join, and spatial joining points to polygons. Is intersect a more reliable way of joining data for finding the census tract ID for each point? It's for the entire USA (92,000 points to 73,000 census tract polygons).
Use the spatial join tool in arctoolbox, you have better control over the inputs and outputs and you can use the Environments tab in the tool to specify other options. Select the intersect option or have their center in... there should be no question whether a point is inside a polygon a polygon, hence my question about erroneous results you were finding.
I would also check to ensure that your points are singlepart points... If you are still finding anomolies, then use the Multipart To Singlepart—Help | ArcGIS for Desktop
on one or maybe even both datasets.
The join that Dan is suggesting offers the possibility to join points to polygons and return the distance too. The distance will be 0 when the point is inside and > 0 when the point is output any polygon.