alternative to else if for random number C#.net

Question

Is there a shorter way to check my random number from 1 - 100 (catNum) against this table of animals? This one doesn't look so bad but I have several more larger tables to work through, I would like to use less lines than I would have to using the statement below:

if (catNum < 36) { category = "Urban"; }
        else if (catNum < 51) { category = "Rural"; }
        else if (catNum < 76) { category = "Wild"; }
        else if (catNum < 86) { category = "Wild Birds"; }
        else { category = "Zoo"; }

Example of further tables:

Answer 1

I prefer to use something like this instead of many if/else A category class

class Category
{
    public int Min { get; set; }
    public int Max { get; set; }
    public string Name { get; set; }
}  

Initialise categories once and fill it with your values

var categories = new List<Category>();

and finally a method to resolve the category

public static string Get(int currentValue)
{
    var last = categories.Last(m => m.Min < currentValue);
    //if the list is ordered 
    //or
    // var last = categories.FirstOrDefault(m => m.Min <= currentValue && m.Max >= currentValue);
    return last?.Name;
}

Answer 2

One alternative is to build up a full list of the items, then you can just select one, at random, by index:

var categories =
    Enumerable.Repeat("Urban", 35)
        .Concat(Enumerable.Repeat("Rural", 15))
        .Concat(Enumerable.Repeat("Wild", 25))
        .Concat(Enumerable.Repeat("Wild Birds", 10))
        .Concat(Enumerable.Repeat("Zoo", 15))
        .ToArray();

var category = categories[45]; //Rural;

Answer 3

Yes, this is a well-studied problem and there are solutions that are more efficient than the if-else chain that you've already discovered. See https://en.wikipedia.org/wiki/Alias_method for the details.

My advice is: construct a generic interface type which represents the probability monad -- say, IDistribution<T>. Then write a discrete distribution implementation that uses the alias method. Encapsulate the mechanism work into the distribution class, and then at the use site, you just have a constructor that lets you make the distribution, and a T Sample() method that gives you an element of the distribution.

I notice that in your example you might have a Bayesian probability, ie, P(Dog | Urban). A probability monad is the ideal mechanism to represent these things because we reformulate P(A|B) as Func<B, IDistribution<A>> So what have we got? We've got a IDistribution<Location>, we've got a function from Location to IDistribution<Animal>, and we then recognize that we put them together via the bind operation on the probability monad. Which means that in C# we can use LINQ. SelectMany is the bind operation on sequences, but it can also be used as the bind operation on any monad!

Now, given that, an exercise: What is the conditioned probability operation in LINQ?

---

Remember the goal is to make the code at the call site look like the operation being performed. If you are logically sampling from a discrete distribution, then have an object that represents discrete distributions and sample from it.

Answer 4

string[] animals = new string[] { "Urban", "Rural", "Wild", "Wild Birds", "Zoo" };
int[] table = new int[] { 35, 50, 75, 85 };     
for (int catNum = 10; catNum <= 100; catNum += 10)
{
    int index = Array.BinarySearch(table, catNum);
    if (index < 0) index = ~index;
    Console.WriteLine(catNum + ": " + animals[index]);
}

Run online: https://dotnetfiddle.net/yMeSPB