Swift function to pseudorandomly choose an item from an Array, allowing for individual weighting of their likelihood of choice.

PickWeighted.swift

/// Pseudorandomly choose a value from an array of values, where each is weighted relative to the others in the collection.
///
/// The `weight` component of each tuple is a positive integer signalling how likely a value is to be chosen vs. other values with different weights. For example, a value with a weight of `4` is four times as likely as a value with a weight of `1` to be chosen.
///
/// - Parameter values: an array of `(value:Thing,weight:Int)` tuples.
///
/// - Returns: a single chosen value.
///
func pickWeighted<Thing>(from weightedValues: [(value: Thing, weight: Int)]) -> Thing {
    // compute the sum of the weights of all of the value/weight pairs
    let totalWeight = weightedValues.map({$0.weight}).reduce(0,+)
    // strategy: pick a random number up to the totalWeight, and then iterate the individual weights until we've exhausted the random value, effectively letting the spinner stop on that value.
    // eg, given [1,3,5,1] a spinnerValue of 6 will land on the `5`.
    var spinnerValue = Int.random(in: 1...totalWeight)
    for weightedValue in weightedValues {
        spinnerValue -= weightedValue.weight
        if spinnerValue <= 0 {
            return weightedValue.value
        }
    }
    fatalError("It should not have been possible to run off the end of totalWeight without landing on a value.")
}

// Test it out!
// let's run it ten thousand times and see how often each value gets chosen.
let inputWeightedValues = [(value: "Hello", weight: 2), // should be around 10%
                           (value: "World", weight: 1), // should be around 5%
                           (value: "Nice", weight: 10), // should be around 50%
                           (value: "Day", weight: 1),   // should be around 5%
                           (value: "Today", weight: 1), // should be around 5%
                           (value: "Huh", weight: 5)]   // should be around 25%
var chosenValueCounts = [String:Int]()
let numberOfTests = 10_000
for _ in 1...numberOfTests {
    let chosenValue = pickWeighted(from: inputWeightedValues)
    chosenValueCounts[chosenValue, default: 0] += 1
}
var chosenValueFrequencies: [String:Double] = chosenValueCounts.mapValues {
    (numberOfTimesChosen: Int) -> Double in
    let frequency = Double(numberOfTimesChosen) / Double(numberOfTests)
    return frequency
}
print(chosenValueFrequencies)