median calculation tool

4927
9
05-03-2016 11:02 AM
GarrettPullis
New Contributor II

Currently working on a Median calculation tool. This tool is similar to the one I posted on a previous post involving selecting a city and zooming into it. In this case, we have to type in a country name, where the tool will then select all the cities in that country, sort the POP_RANK of the cities in that country, and calculate the median POP_RANK for that country. I was able to complete the select and zoom aspect but I am not entirely sure how to create a list/search cursor which will allow us to input a calculation. Is there a specific way I should approach this?  Any and all help is appreciated!! Thank you

import arcpy

from arcpy import mapping

#sets the workspace

mxd = arcpy.mapping.MapDocument("CURRENT")

df = arcpy.mapping.ListDataFrames(mxd, "Layers")[0]

fc = ("N:/data2016/Lab13/cities.shp")

cities = arcpy.mapping.ListLayers(mxd, "Cities")[0]

cities.showLabels = True

def countryList():

    country = arcpy.GetParameterAsText(0)

    country_Layer = arcpy.MakeFeatureLayer_management(fc, "country_lyr")

    arcpy.SelectLayerByAttribute_management(cities, "NEW_SELECTION", "CNTRY_NAME = '{}' ".format(country))

    arcpy.mapping.ListDataFrames(mxd)[0].zoomToSelectedFeatures()

    arcpy.RefreshActiveView()

    arcpy.Delete_management("country_lyr")

0 Kudos
9 Replies
DanPatterson_Retired
MVP Emeritus

SearchCursor—Help | ArcGIS for Desktop example 2 shows how to ge the unique values.. if you leave out the set(...) bit, you have all the values, which then you can then use a python sort, get the len(gth) and if len is divisible by 2 (mod), just take the average of the middle 2 (from len/2 and len/2 +1) otherwise if len is odd, just take the middle one.

alternately that field can be read and converted to a numpy array for which there are median methods for data with and without nodata values.

DanPatterson_Retired
MVP Emeritus

forgot the code link http://www.arcgis.com/home/item.html?id=6c384f06c9f14d09920f4ff14460f4e2

and your chance to join a cause to free statistical tools Free Frequency ... join the cause

DanPatterson_Retired
MVP Emeritus

If you install ArcGIS Pro, you will be using python 3.4 which comes with the Statistics module, which has 4 variants of the median which covers most useable situations 9.7. statistics — Mathematical statistics functions — Python 3.4.4 documentation

0 Kudos
DanPatterson_Retired
MVP Emeritus

Does anyone in the 'global reach' section have any further ideas? or should I mark this 'assumed answered'

0 Kudos
NeilAyres
MVP Alum

There was a median calculation tool posted here somewhere by Caleb Mackey.

But no joy with search. So here it is again.

'''
Written By Caleb Mackey
4/17/2013
Calculates Median Statistics
'''
import arcpy, os, sys, traceback
# env settings
arcpy.env.overwriteOutput = True
arcpy.env.qualifiedFieldNames = False
def GetMedian(in_list):
    sorted_list = sorted(in_list)
    median = int(round(len(sorted_list) / 2))
    if len(sorted_list)%2==0:
        med_val = float(sorted_list[median-1]
                        + sorted_list[median]) / 2
    else:
        med_val = sorted_list[median]
    return med_val
def GetMedianValues(source_fc, new_table, case_field, value_field):
    
    ''' Generates a table with Median Values, summarized by case_field. If the
        goal is to get the median for the entire table, use a case field that has
        the same value for all records.
        source_fc - input feature class to compute median statistics for
        new_table - output table
        case_field - similar to dissolve field, computes stats based on unique values in this field
        value_field - field that contains the actual values for statistics; must be numeric
    '''
    
    # Get unique value list for query
    print 'starting cursor'
    with arcpy.da.SearchCursor(source_fc, [case_field]) as rows:
        un_vals = list(set(r[0] for r in rows))
    lyr = arcpy.MakeFeatureLayer_management(source_fc,'source_layer')
    values = {}
    # Get Median UseValue for each station name
    for st in un_vals:
        query = '"{0}" = \'{1}\''.format(case_field, st)
        arcpy.SelectLayerByAttribute_management(lyr, 'NEW_SELECTION', query)
        use_vals = []
        with arcpy.da.SearchCursor(lyr, [value_field]) as rows:
            for row in rows:
                if row[0] != None:
                    use_vals.append(row[0])
        if len(use_vals) > 0:
            median = GetMedian(use_vals)
            values[st] = [median, len(use_vals)]
    # Create new Summary Statistics table with median
    #
    if arcpy.Exists(new_table):
        arcpy.Delete_management(new_table)
    arcpy.CreateTable_management(os.path.split(new_table)[0],os.path.basename(new_table))
    # Get field names and types
    for field in arcpy.ListFields(source_fc):
        if field.name in [case_field, value_field]:
            ftype = field.type
            name = field.name
            length = field.length
            pres = field.precision
            scale = field.scale
            if name == value_field:
                if new_table.endswith('.dbf'):
                    name = 'MED_' + value_field[:6]
                else:
                    name = 'MED_' + value_field
                value_field2 = name
            arcpy.AddField_management(new_table,name,ftype,pres,scale,length)
            
    # Add frequency field
    arcpy.AddField_management(new_table,'FREQUENCY','LONG')
    # Insert rows
    with arcpy.da.InsertCursor(new_table, [case_field, value_field2, 'FREQUENCY']) as rows:
        for k,v in sorted(values.iteritems()):
            rows.insertRow((k, v[0], v[1]))
            
    # report results
    print 'Created %s' %os.path.basename(new_table)
    arcpy.AddMessage('Created %s' %os.path.basename(new_table))
    # .dbf's are automatically given a 'Field1' field...Clean this up
    try:
        if new_table.endswith('.dbf'):
            arcpy.DeleteField_management(new_table, 'Field1')
    except:
        pass
    print 'Done'
if __name__ == '__main__':
##    # testing
##    source_fc = r'C:\Testing\Test.gdb\CSR_by_TWP'
####    new_table = r'C:\Testing\Test.gdb\Median_CSR' #gdb test
##    new_table = r'C:\Testing\Median_CSR.dbf'  #dbf test
##    case_field = 'NAME'
##    value_field = 'AVE_CSR'
    # Script tool params
    source_fc = arcpy.GetParameterAsText(0)
    new_table = arcpy.GetParameterAsText(1)
    case_field = arcpy.GetParameterAsText(2)
    value_field = arcpy.GetParameterAsText(3)
    GetMedianValues(source_fc, new_table, case_field, value_field)
DanPatterson_Retired
MVP Emeritus

so if using recent versions of python, it would still be easier to use the builtins

>>> import statistics

>>> a = [1,2,3]

>>> statistics.median(a)

2

>>> b = [1,2,3,4]

>>> statistics.median(b)

2.5

>>>

>>> dir(statistics)

['Decimal', 'Fraction', 'StatisticsError', '__all__', '__builtins__', '__cached__', '__doc__', '__file__', '__loader__', '__name__', '__package__', '__spec__', '_check_type', '_counts', '_decimal_to_ratio', '_exact_ratio', '_ss', '_sum', 'collections', 'math', 'mean', 'median', 'median_grouped', 'median_high', 'median_low', 'mode', 'pstdev', 'pvariance', 'stdev', 'variance']

>>>

NeilAyres
MVP Alum

Of course Dan,

But I am running 10.3.1 (soon moving to 10.4).

Haven't got anywhere near Pro yet. Need to buy a deep blue before I do.

0 Kudos
DanPatterson_Retired
MVP Emeritus

I have been using python 3.4 since arcmap 10.2... it is a matter of setup

0 Kudos
DanPatterson_Retired
MVP Emeritus

I think we have it wrapped up with this one, in case it is a network question

image.gif