ee.FeatureCollection.randomColumn

AI-generated Key Takeaways

• randomColumn() adds a new column of pseudorandom numbers to a FeatureCollection, with the default column name being "random".

• The generated random numbers can follow either a uniform distribution ([0, 1)) or a normal distribution (μ=0, σ=1) specified using the distribution parameter.

• Users can provide a seed value for reproducibility using the seed parameter, ensuring the same sequence of random numbers is generated for a given seed.

• This function is commonly used for tasks like randomly splitting a FeatureCollection into subsets for training and testing machine learning models, as demonstrated in the examples.

Adds a column of deterministic pseudorandom numbers to a collection. The outputs are double-precision floating point numbers. When using the 'uniform' distribution (default), outputs are in the range of [0, 1). Using the 'normal' distribution, outputs have μ=0, σ=1, but have no explicit limits.

Usage	Returns
FeatureCollection.randomColumn(columnName, seed, distribution, rowKeys)	FeatureCollection

Argument	Type	Details
this: collection	FeatureCollection	The input collection to which to add a random column.
columnName	String, default: "random"	The name of the column to add.
seed	Long, default: 0	A seed used when generating the random numbers.
distribution	String, default: "uniform"	The distribution type of random numbers to produce; one of 'uniform' or 'normal'.
rowKeys	List, optional	A list of properties that should uniquely and repeatably identify an element of the collection, used to generate the random number. Defaults to [system:index].

Examples

Code Editor (JavaScript)

// FeatureCollection of power plants in Belgium.
var fc = ee.FeatureCollection('WRI/GPPD/power_plants')
            .filter('country_lg == "Belgium"');
print('N features in collection', fc.size());

// Add a uniform distribution random value column to the FeatureCollection.
fc = fc.randomColumn();

// Randomly split the collection into two sets, 30% and 70% of the total.
var randomSample30 = fc.filter('random < 0.3');
print('N features in 30% sample', randomSample30.size());

var randomSample70 = fc.filter('random >= 0.3');
print('N features in 70% sample', randomSample70.size());

Python setup

See the Python Environment page for information on the Python API and using geemap for interactive development.

import ee
import geemap.core as geemap

Colab (Python)

# FeatureCollection of power plants in Belgium.
fc = ee.FeatureCollection('WRI/GPPD/power_plants').filter(
    'country_lg == "Belgium"')
print('N features in collection:', fc.size().getInfo())

# Add a uniform distribution random value column to the FeatureCollection.
fc = fc.randomColumn()

# Randomly split the collection into two sets, 30% and 70% of the total.
random_sample_30 = fc.filter('random < 0.3')
print('N features in 30% sample:', random_sample_30.size().getInfo())

random_sample_70 = fc.filter('random >= 0.3')
print('N features in 70% sample:', random_sample_70.size().getInfo())