WeightedRandom - Select content based on weights (C# / .NET)

Need to select content from a collection based on weights?  Here is a great .NET 4+ class that is very helpful.

The class can be used with just about any collection and you can pass in a delegate/lamda to determine what the weight is on each collection object.

	using System;
	using System.Collections.Generic;
	using System.Linq;
	using System.Text;
	using System.Threading.Tasks;
	//
	// THIS IS ALL JUST TEST CODE
	//
	// THE ACTUAL WEIGHTED RANDOM CODE IS IN THE WeightedRandom.cs file
	//
	namespace PaulTechGuy.Weights
	{
	class Program
	{
	class Foo
	{
	public string Name { get; set; }
	public int Weight { get; set; }
	public Foo(string name, int weight)
	{
	Name = name;
	Weight = weight;
	}
	}
	static void Main(string[] args)
	{
	new Program().Run(args);
	}
	private void Run(string[] args)
	{
	// test data
	Foo[] items = new Foo[]
	{
	// there are six individual weights from 10 to 60)
	new Foo("Sally Target=17% (1/6th)", 10),
	new Foo("Fred Target=30% (1/3th)", 30), // or 2/6
	new Foo("John Target=50% (1/2th)", 60), // or 3/6
	};
	// here is the magic; declare your collection Func weight delegate and
	// create the WeightedRandom object; after that you just call Next() to
	// get your next random object based on weights
	Func<Foo, int> fweight = f => f.Weight;
	WeightedRandom<Foo> foos = new WeightedRandom<Foo>(items, fweight, new Random());
	// spit out a bunch of test results
	for (int times = 0; times < 5; ++times)
	{
	// generate a bunch of random test data and store counts
	Dictionary<string, int> statistics = new Dictionary<string, int>();
	for (int i = 0; i < 1000; ++i)
	{
	Foo foo = foos.Next();
	if (!statistics.ContainsKey(foo.Name))
	{
	statistics.Add(foo.Name, 0);
	}
	statistics[foo.Name] += 1;
	}
	// print out results
	double total = statistics.Sum(s => s.Value);
	foreach (var kvp in statistics)
	{
	Console.WriteLine("{0} => {1} (Actual {2}%)", kvp.Key, kvp.Value, Math.Round(kvp.Value / total * 100));
	}
	Console.WriteLine();
	}
	}
	}
	}

view raw Program.cs hosted with ❤ by GitHub

	using System;
	using System.Collections.Generic;
	using System.Diagnostics.Contracts;
	using System.Linq;
	namespace PaulTechGuy.Weights
	{
	public class WeightedRandom<T>
	{
	private readonly List<IGrouping<int, T>> _itemMap;
	private readonly Random _random;
	private readonly int _maxWeight;
	public WeightedRandom(T[] items, Func<T, int> weight, Random rnd)
	{
	Contract.Requires(items.Length > 0);
	_random = rnd;
	// map the items by weight and sort by weight
	_itemMap = items
	.GroupBy(weight)
	.OrderBy(g => weight)
	.ToList();
	// weights are sorted in ascending order; max weight will be last element
	_maxWeight = _itemMap.ElementAt(_itemMap.Count - 1).Key;
	}
	public T Next()
	{
	// select a random weight
	int randomWeight = _random.Next(_maxWeight + 1);
	// get the first group where our random weight is less then element weight
	IGrouping<int, T> items = _itemMap
	.FirstOrDefault(b => randomWeight <= b.Key);
	// since we based our randomWeight on the maxWeight, we better have found one
	if (items == null)
	{
	throw new InvalidOperationException("Internal bug: should have found at least the maxWeight element");
	}
	// there may be more than one element with this weight; take a random one
	T item = items.ElementAt(_random.Next(items.Count()));
	return item;
	}
	}
	}

view raw WeightedRandom.cs hosted with ❤ by GitHub