NEED HELP IMMINENTLY WITH MULTIVARIATE ANALYSIS TOOL

[NicoleGans]

I am using the tool "multivariate Analysis" 

We are using an feature layer that is 4 intersecting layers. for the Initialization Method, we selected "user defined field" and then created a unique column in the input feature. This column was used as the initialization field. 

We then got the following errors: 

WARNING 000916: The input feature class does not appear to contain projected data.
WARNING 000642: Problems reading 4 of 317581 total records.
WARNING 000848: Features with bad records (only includes first 30): OBJECTID = 57162, 205863, 252995, 275999.
ERROR 110130: The number of User Defined Seed Locations must be larger than one and smaller than the number of total features.
Failed to execute (MultivariateClustering).

Can you please let us know how to proceed. We have exhausted our troubleshooting knowledge. 

[DanPatterson_Retired]

You have bad geometry and the tool appears to require projected data.

1  run the Repair Geometry tool from arctoolbox.

2  with the remaining good geometry, use the Project tool and project your data to a new file with an appropriate projected coordinate system

[NicoleGans]

Hi Dan, 

I took the steps you mentioned, and I had the exact same errors. 

1) For the Project tool, we just used the output coordinate system of the input dataset? Is there something else we should be using? 

2) For the repair geometry, we used the ESRI validation method. 

Thanks for your help. 

[NicoleGans]

Hi -- Sorry to bother you again. The bigger issue we believe is with our seed. What does the initialization field need to be? The error is "ERROR 110130: The number of User Defined Seed Locations must be larger than one and smaller than the number of total features."

How can we create a column to do this? 

[DanPatterson_Retired]

Nicole, perhaps if you provide a direct link to the help topic you are using and the version of the software, someone may be able to help.  There is no 'multivariate analysis' in Pro, but lots of multivariate tools.

As for the last issue, the process apparently wants at least 2 'seed' locations and no more than the number of points in the dataset.  It remains to be seen whether you have to select records within the point set or whether they have to be two or more points in a different featureclass.

As for the Projection question... a projection is when the data coordinates have been changed to planar coordinates from spherical coordinates.  If your input dataset does not have a 'projection' (eg. UTM, Albers etc) then it hasn't been projected.

When repair geometry was done, you should now have no more invalid geometries, which is good... you just don't get a message to that affect.