Weighted Random Distribution

A simple utility module to select from a set of discrete values based on weighted random distribution.

Table of Contents

1. inPlaceUnsorted()
2. normalize()

inPlaceUnsorted()

Perform an in-place unsorted weighted value selection. (The algorithm is ported from Python code here).

Parameters

• weights {object} - An Object with each key representing one of the discrete values that will be selected, and each value representing a weight.

Returns

One of the keys from the weights param, with the probability of a key being returned based on the weight values {string}

Example

inPlaceUnsorted({ 'a': 7, 'b': 2, 'c': 1 })
// output: 70% chance of 'a', 20% chance of 'b', 10 percent chance of 'c'
 
inPlaceUnsorted({ 'a': .1, 'b': .05, 'c': .05 })
// output: 50% change of 'a', 25% chance of 'b', 25% chance 'c'

Numeric values don't need to sum to any particular number, and can be decimal.

normalize()

A utility method to provide min-max normalization for input weights. Given a starting range of xMin and xMax, returns valueToNormalize mapped to the same relative position in relation to transformedMin and transformedMax. Defaults to a transformed range between 0 and 1.

Parameters

• valueToNormalize {number}
• xMin {number} The floor of the normalization range.
• xMax {number} The ceiling of the normalization range.
• transformedMin {number} Default value is 0.
• transformedMax {number} Default value is 1.

Returns

The normalized representation of valueToNormalize {number}.

Examples

normalize(50, 0, 100);
// output: .5
 
normalize(50, 0, 100, 0, 8);
// output: 4
 
normalize(20, 10, 20);
// output: 1
 
normalize(10, 10, 20, 30, 40);
// output: 30
 
normalize(20, 10, 20, 30, 40);
// output: 40