Jenness Enterprises
Wildlife and Forestry Tools for ArcGIS Pro — Tool Reference
jennessent.com
Tool Reference › Data Tools › Add Random Numbers

Add Random Numbers

Data Tools · geoprocessing tool · by Jeff Jenness and Matt Wiebers
Works at every ArcGIS Pro license level

Summary

Fills a numeric field of a layer's or table's records — only the selected records when there is a selection, and only the records a definition query shows — with a constant, a random permutation of 1 to N (every record a different number: the classic way to assign treatments to plots at random), the same permutation repeated for each replicate (a case field groups the records, and every group gets its own 1 to n), random integers within a range, random decimals between 0 and 1 or within a range, or draws from a normal distribution with a given mean and standard deviation. A field that does not exist yet is created, as an integer or double-precision field to suit the values, and a seed makes the result repeatable.

Add Random Numbers is a port of the tool of the same name in the CropCentric ArcGIS tools that Jeff Jenness built with Matt Wiebers for ArcMap (Jenness and Wiebers 2018), which wrote a constant or a random permutation; the repeated permutation, the ranges, the unit interval and the normal distribution are additions.

Usage

Which records. A map layer or table view writes random numbers to only the records it sees, so setting a definition query on a layer will mean only those records get the new values. Selections are also honored, so only selected records will be modified. If you browse for a dataset on the hard drive, then all records will automatically be calculated, because datasets on the hard drive have no definition query or selection. N, for the permutation, is the number of records written; for the repeated permutation it is the size of each group.

The Data Tools group of the ribbon, with the Add Random Numbers button outlined in blue
Where to find it: Add Random Numbers is in the Data Tools group of the Wildlife and Forestry tab.

The values.

How the numbers are made. Every mode draws from Python's standard random module, seeded with your seed, so the same seed with the same records reproduces any mode exactly. The uniform numbers underneath come from the Mersenne Twister (Matsumoto and Nishimura 1998), with 53 random bits per value. The permutation modes are Fisher–Yates shuffles in Durstenfeld's form (Fisher and Yates 1938; Durstenfeld 1964; Knuth 1997, Algorithm P), each index drawn by rejection sampling, so every ordering is equally likely and there is no modulo bias; the integer range is drawn the same way. Random numbers generated from a Normal distribution use the Box–Muller transform in its original trigonometric form (Box and Muller 1958), not the polar variant: from two uniform values U1 and U2 on [0, 1) it forms

z1=cos⁡(2πU1)−2⁢ln⁡(1−U2),z2=sin⁡(2πU1)−2⁢ln⁡(1−U2),

two independent standard-normal values, which serve two successive records and are scaled by the mean and standard deviation you give. The run messages and the field's metadata record the method used, so a methods section can name it.

The field. The field list offers the input's writable numeric fields (the ObjectID, the geometry and read-only measures such as Shape_Length are left out); pick an existing field to write the new values in place, or type a name of your own. Integer fields take only the integer modes and whole-number constants. The dialog will stop you if you try to write a decimal number, or a constant with a decimal part, into an integer field. A name that does not exist yet is created as a LONG field for integer values or a DOUBLE field for decimals, shortened to ten characters on a shapefile or dBASE table. The default name, Treatment, matches the empty field that Create Array along Axes leaves for it.

Repeatability. Give a seed to get the same numbers again; leave it empty for a fresh draw each run. The run is recorded in the geoprocessing History either way.

A worked example

We used Plot Designer to lay out our plots: four replicates of a 2 by 6 design, the replicates numbered 1 through 4 and the plots within each replicate numbered 1 through 12, 48 plots in all.

An aerial photograph of a field covered by four colored blocks of twelve plots each, every block numbered 1 to 12 in the same order from its southern edge upward
The 48 plots as Plot Designer made them: four replicates (the colors), each numbered 1 to 12 in the same order.

We want to test the plots with some treatment and use the plot numbers to identify the treatments, but the current arrangement is likely to skew the results just because of the way the field itself varies over the landscape. The effects we see from treatments 1 and 2, for example, might be due to the fact that they are all on the southern edge of the field. We want to scramble those treatment numbers within each replicate, so we run the tool with Random permutation of 1 to N (repeated for replicates), writing into the Plot field itself, with the replicate number as the case field. The field list shows that field by its alias, Replication.

The Add Random Numbers dialog: input sample_boundary_Plots, field Plot, values Random permutation of 1 to N (repeated for replicates), case field Replication, no seed
The dialog: the Plot field, the permutation repeated for replicates, and Replication as the case field.

Every replicate still holds each number from 1 to 12 once, but in its own random order, so no treatment is tied to one edge of the field:

The same four colored blocks of twelve plots, now numbered in a different random order within each block, every block still showing each number from 1 to 12 once
After the run: 1 to 12 dealt afresh within each of the four replicates.

If we want to give a random unique ID number to every plot instead, we use Random permutation of 1 to N (each record unique), which needs no case field:

The Add Random Numbers dialog with field Plot and values Random permutation of 1 to N (each record unique); the case field is not shown
The dialog for one permutation over all 48 plots.
The four colored blocks of plots numbered from 1 to 48 in random order across the whole field
After the run: every plot has its own number from 1 to 48, in random order across the whole field.

The tool will also generate random numbers from a normal distribution, and the other kinds described above, which you might want for many statistical and analytical purposes.

In ModelBuilder the tool takes the layer in and hands the same layer on, updated, for the next step:

A ModelBuilder diagram: the sample_boundary_Plots layer, the Add Random Numbers tool, and the updated sample_boundary_Plots (2) output
Add Random Numbers in a model: the input layer in, the updated layer out.

Parameters

LabelExplanationData type
Input table or layerRequired · in_table The layer, table view, feature class or table whose records are written. A layer's selection and definition query are honored. Table View
Field to fillRequired · field Pick one of the input's writable numeric fields to fill it in place, or type a new field name, created to suit the values. An integer field refuses decimal modes and constants with a decimal part. String
Values to writeRequired · mode Constant value; Random permutation of 1 to N (each record unique); Random permutation of 1 to N (repeated for replicates); Random integers within a range; Random decimals between 0 and 1; Random decimals within a range; Random decimals from a normal distribution (mean, standard deviation). String
Case fieldOptional · case_field For the repeated-permutation mode: the field whose values identify the replicates. Each distinct value is dealt its own permutation of 1 to n, n being the number of records with that value. Suggested as Rep when the input has that field. Field
Constant valueOptional · constant The value for the constant mode. Double
Lowest value, highest value of the rangeOptional · low, high The range for the two range modes; inclusive for integers. Double
Mean, standard deviationOptional · mean, sd The normal distribution's parameters; the standard deviation cannot be negative. Double
Random seedOptional · seed Any whole number. The same seed with the same records gives the same numbers again. Long
Updated tableDerived · out_table The input, after writing, for chaining in ModelBuilder. Table View

Python

Deal the treatment numbers 1 to n afresh within every replicate of a Plot Designer output, whose Rep field is the case field:

import arcpy
arcpy.ImportToolbox(r"C:\path\to\JennessEnterprisesTools.pyt")  # your install path
arcpy.jenness.AddRandomNumbers(
    in_table=r"D:\data\trial.gdb\Plots_2026", field="Treatment",
    mode="Random permutation of 1 to N (repeated for replicates)",
    case_field="Rep", seed=42)

Deal 1 to N to the plots of replication 2 alone, in random order:

arcpy.management.MakeFeatureLayer(r"D:\data\trial.gdb\Plots_2026", "plots")
arcpy.management.SelectLayerByAttribute("plots", "NEW_SELECTION", "Rep = 2")
arcpy.jenness.AddRandomNumbers(
    in_table="plots", field="Treatment",
    mode="Random permutation of 1 to N (each record unique)", seed=42)

Recommended citation

Jenness, J., and M. Wiebers. 2026. Add Random Numbers. Wildlife and Forestry Tools add-in for ArcGIS Pro, v. 1.113.62.1 (October 2026). Jenness Enterprises. Available at: https://github.com/JeffJenness/Wildlife_Tools.

Credits and references

By Jeff Jenness and Matt Wiebers, Jenness Enterprises (www.jennessent.com). A port of the Add Random Numbers tool of the CropCentric ArcGIS tools for ArcMap (Jenness and Wiebers 2018), with the range, unit-interval and normal-distribution modes added. The random numbers come from Python's standard random module: Mersenne Twister uniforms (Matsumoto and Nishimura 1998), Fisher–Yates shuffles (Fisher and Yates 1938; Durstenfeld 1964; Knuth 1997) and the Box–Muller transform (Box and Muller 1958).

The basemap under the maps in the figures on this page is Esri's World Imagery basemap (source: Esri, Vantor, Earthstar Geographics, and the GIS User Community).

Licensing information

Works at every ArcGIS Pro license level (Basic, Standard, Advanced). No extension licenses are required.