Remove sub-arrays from a numpy array

Question

I'm trying to get a 2D array from a feature class using numpy and arcpy...

import arcpy
import numpy
locxyarray = arcpy.da.FeatureClassToNumPyArray("Points", ['SHAPE@XY', 'PrimeKey'])

The result:

array([([506173.7478, 5455684.263900001], 1),
       ([506175.22869999986, 5455648.723099999], 2),
       ([506229.03359999973, 5455661.5572999995], 3),
       ([506250.25939999986, 5455614.169500001], 4),
       ([506305.54509999976, 5455579.122300001], 5),
       ([506331.70710000023, 5455688.2129], 6)], 
      dtype=[('SHAPE@X', '<f8', (2,)), ('PrimeKey', '<i4')])

What I want:

array([(506173.7478, 5455684.263900001, 1),
       (506175.22869999986, 5455648.723099999, 2),
       (506229.03359999973, 5455661.5572999995, 3),
       (506250.25939999986, 5455614.169500001, 4),
       (506305.54509999976, 5455579.122300001, 5),
       (506331.70710000023, 5455688.2129, 6)], 
      dtype=[('SHAPE@XY', '<f8', (2,)), ('PrimeKey', '<i4')])

I want the above so I can sort by either the 0 or the 1 column, and return the 2 column in that order... Also, I need to be able to calculate the mean of X and the mean of Y, and pull all the Prime Keys with values above or below the mean.

EDIT - now able to make the array "look" correct;

locxyarray = arcpy.da.FeatureClassToNumPyArray("Points", ['SHAPE@X', 'SHAPE@Y', 'PrimeKey'])

>>array([(506173.7478, 5455684.263900001, 1),
       (506175.22869999986, 5455648.723099999, 2),
       (506229.03359999973, 5455661.5572999995, 3),
       (506250.25939999986, 5455614.169500001, 4),
       (506305.54509999976, 5455579.122300001, 5),
       (506331.70710000023, 5455688.2129, 6)], 
      dtype=[('SHAPE@X', '<f8'), ('SHAPE@Y', '<f8'), ('PrimeKey', '<i4')])

But I can't calculate the mean on the X or Y columns... (IndexError: too many indices)

SOLUTION (see comments below in solution, this is a summary):

import arcpy
import numpy as np
locxyarray = arcpy.da.FeatureClassToNumPyArray("Points", ['SHAPE@X','SHAPE@Y','PrimeKey'])
LSideLocs = np.where(locxyarray['SHAPE@X']<=np.mean(locxyarray['SHAPE@X']))
RSideLocs = np.where(locxyarray['SHAPE@X']>=np.mean(locxyarray['SHAPE@X']))

@Cameron don't you want something like np.mean(x['SHAPE@X']) ? — Dalek
– Dalek, Commented Sep 22, 2014 at 21:20

Dalek · Accepted Answer · 2014-09-19 18:25:15Z

2

You can play around with the name of field.

r=[(x['SHAPE@XY'][i][0],x['SHAPE@XY'][i][1],x['PrimeKey'][i]) for i in range(x.shape[0])]
x=np.rec.fromrecords(r, formats = 'f8,f8,i4', names = 'SHAPE@X,SHAPE@Y,PrimeKey')
rec.array([(506173.7478, 5455684.263900001, 1),
       (506175.22869999986, 5455648.723099999, 2),
       (506229.03359999973, 5455661.5572999995, 3),
       (506250.25939999986, 5455614.169500001, 4),
       (506305.54509999976, 5455579.122300001, 5),
       (506331.70710000023, 5455688.2129, 6)], 
      dtype=[('SHAPE@X', '<f8'), ('SHAPE@Y', '<f8'), ('PrimeKey', '<i4')])

answered Sep 19, 2014 at 18:25

Dalek

4,38812 gold badges54 silver badges105 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

Cameron Over a year ago

I was able to split out the SHAPE@X and SHAPE@Y directly in my call line... locxyarray = arcpy.da.FeatureClassToNumPyArray("Points", ['SHAPE@X', 'SHAPE@Y', 'PrimeKey']) But the resulting array still has too many indices to be able to calculate the mean of the X or Y column... (after that I want to pull the PrimeKeys of the rows above and below the mean to be able to split the table by X or Y)

Cameron Over a year ago

I realised it was a bit too vague, so I tried to provide more detail above... essentially I need to be able to sort the array based on X and/or Y, and grab the corresponding PrimeKey's in the order I want (i.e. either from north to south, or east to west). I also need to be able to split the values in half (i.e. grab the east half, and order from north to south, then grab the west half, and order from north to south) I'm working to build labels that located regularly in layout space, but are leadered and orderd from north to south based on their data frame coordinates...

Cameron Over a year ago

That looks promising... first I need the mean, then too be able to split the array between values above the mean and values below (i.e. split the labels between the right and left half of the page), then sorting based on Y would be perfect (let's me label from north to south). I'm struggling to make it work though... if I take you're suggestion of "np.mean(x['SHAPE@X'])" (or xmean = locxyarray.mean(x['SHAPE@X'])) then I get errors that it's an invalid index... If i try to argsort, I get an error that an integer is required...

Dalek Over a year ago

@Cameropn have you tried this:np.mean(locxyarray['SHAPE@X'])? Do you get error with it?

Dalek Over a year ago

@Cameron you can use np.where(locxyarray['SHAPE@X']<=np.mean(locxyarray['SHAPE@X'])) to again get the indices of elements less or equal to mean value.

|

Anton Savin · Accepted Answer · 2014-09-19 18:13:20Z

0

Something like this:

>>> numpy.array([(x[0][0], x[0][1], x[1]) for x in locxyarray], dtype=np.dtype([('X', np.float), ('Y', np.float), ('PrimeKey', np.int32)]))
array([(506173.7478, 5455684.263900001, 1),
   (506175.22869999986, 5455648.723099999, 2),
   (506229.03359999973, 5455661.5572999995, 3),
   (506250.25939999986, 5455614.169500001, 4),
   (506305.54509999976, 5455579.122300001, 5),
   (506331.70710000023, 5455688.2129, 6)],
  dtype=[('X', '<f8'), ('Y', '<f8'), ('PrimeKey', '<i4')])

Note that dtype can't be kept the same because now the array elements consist of three fields instead of two in the original.

answered Sep 19, 2014 at 18:13

Anton Savin

41.4k8 gold badges57 silver badges93 bronze badges

Collectives™ on Stack Overflow

Remove sub-arrays from a numpy array

2 Answers 2

6 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

6 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related