Pull one attribute from a featureclass

256
9
01-22-2020 11:02 PM
QuangTruong
New Contributor II

I am a beginner, and the following question will show why. 

I am trying to extract one value from a layer that has only one feature in it, but the only way I know how to do it takes three lines of code. There must be a better way, right? See the below code that I am using, and excuse the ignorance. 

# pull some attributes from the isolated feature
with arcpy.da.SearchCursor(fcTemp, 'ATTR') as pulllist:
    for pull in pulllist:
        var = pull[0]
0 Kudos
9 Replies
DanPatterson_Retired
MVP Esteemed Contributor

10,000 records, 

basically line 1 doesn't count... I could have put line 2 into line 4 and the rest is viewing fluff.

So ignoring blank lines, imports and putting file names on separate lines... line 4 is an data access module incarnation of a searchcursor(ish).  The rest is numpy.  You can work with numpy (line 6) or pandas... you can view/use it like a list.

import arcpy
in_tbl = r"c:\somepath\some.gdb\sometable"

d = arcpy.da.TableToNumPyArray(in_tbl, "Age")  # pull all the `age` data in

d0 = d.view(np.int32)
d0
array([61, 31, 60, 49, 85, 55, 39, ..., 71, 38, 44, 38, 21, 28, 88])

d1 = d0.tolist()
[61, 31, 60, 49, 85, 55, 39, ... huge snip ... , 71, 38, 44, 38, 21, 28, 88]‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

If you just wanted the first value.. you slice line 4 d[0][0]

out = []
with arcpy.da.SearchCursor(in_tbl, 'Age') as pulllist:
    for pull in pulllist:
        out.append(pull[0])
        
out[:10]  # ---- slice first 10
[61, 31, 60, 49, 85, 55, 39, 20, 63, 19]‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Timings are comparable if things have been called at least once

%timeit arcpy.da.TableToNumPyArray(in_tbl, "Age")  # pull all the `age` data in
9.38 ms ± 398 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

%%timeit
out = []
with arcpy.da.SearchCursor(in_tbl, 'Age') as pulllist:
    for pull in pulllist:
        out.append(pull[0])
        
9.94 ms ± 311 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
QuangTruong
New Contributor II

Also, I'm interested in how you're timing the functions. 

Can you tell me more about how to measure the time it takes to run not just one command, but my entire script? 

0 Kudos
DanPatterson_Retired
MVP Esteemed Contributor

many ways... but just add these to your script

from datetime import datetime
from time import sleep
startTime = datetime.now()


sleep(0.5)  #  put your code here instead of sleep


print(datetime.now() - startTime)
0:00:00.510107

0 Kudos
JakeSkinner
Esri Esteemed Contributor

Hi Quang,

Here is an example of your code consolidated into 1 line:

var = [row for row in arcpy.da.SearchCursor(fcTemp, 'ATTR')][0][0]
DanPatterson_Retired
MVP Esteemed Contributor
d = arcpy.da.TableToNumPyArray(in_tbl, "Age")[0][0]# ---- or in your case
var = arcpy.da.TableToNumPyArray(fcTemp, "ATTR")[0][0]‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

 which are probably the same under their shells

But

# --- you can get frequency counts
#
h = np.histogram(arcpy.da.TableToNumPyArray(in_tbl, "Age")['Age'])

h # ---- top row counts, bottom bin counts
(array([1495, 1297, 1617, 1710, 1185,  801,  658,  407,  348,  482],
       dtype=int64),
 array([18. , 25.1, 32.2, 39.3, 46.4, 53.5, 60.6, 67.7, 74.8, 81.9, 89. ]))‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Or derive statistics that you can't do with searchcursors

v = arcpy.da.TableToNumPyArray(in_tbl, "Age")['Age']

# ---- like the median

v.min(), v.max(), v.std(), np.median(v)

(18, 89, 17.688446350937664, 42.0)‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Esri provides the tools, but not everyone uses them

Even access to graphing from the comfort of your favorite IDE or you can do it in ArcGIS Pro and set things and get the statistics there

import matplotlib.mlab as mlab
import matplotlib.pyplot as plt

num_bins = 5

n, bins, patches = plt.hist(v, num_bins, facecolor='blue', alpha=0.5)‍‍‍‍‍‍‍‍‍‍‍‍

JoshuaBixby
MVP Esteemed Contributor

If you only want to retrieve the first record of a cursor:

var, = next(arcpy.da.SearchCursor(fcTemp, 'ATTR')) # note comma for tuple unpacking
DanPatterson_Retired
MVP Esteemed Contributor

Love a good lesson session

How about every second record?

or the data sorted?

arcpy.da.TableToNumPyArray(in_tbl,"Age")["Age"]
array([61, 31, 60, ..., 21, 28, 88])

arcpy.da.TableToNumPyArray(in_tbl,"Age")["Age"][::2]
array([61, 60, 85, ..., 38, 38, 28])

np.sort(arcpy.da.TableToNumPyArray(in_tbl,"Age")["Age"])
array([18, 18, 18, ..., 89, 89, 89])
QuangTruong
New Contributor II

These are all super helpful! Thank you guys!

JoshuaBixby
MVP Esteemed Contributor

Please mark one of the responses as Correct or mark the question as "Assumed answer" to close it out.

0 Kudos