Tag Archives: Probability

Will it Rain this Weekend?

While browsing MIT+K12 open projects, I came across an entire course proposal about probability. Now, I have very little formal probability training outside of one unit in precalculus last year, but I was intrigued by the idea of writing a program to simulate 10,000 trials. Obviously, 10,000 trials isn’t enough to get a really good data set so I added a few zeros to my C# for loop and ran 1,000,000 trials.

So, in keeping with the MIT course proposal, I checked the forecast for the weekend and found out that this weekend, the chance of rain on Saturday is 10% and the chance of rain on Sunday is 30%. My first reaction was mild aggravation because I went camping last weekend in the rain and I simply could’ve waited a week. My second reaction was, “Alright. Let’s get started.”

Probability Tree Image

Mr. Whalen, I hope you’re proud of me.

The next step was to write a program to test my calculations. I came up with a program structure that I hope is “random” enough. I am going to use System.Random instead of RNGCryptoServiceProvider because it is way faster and I believe it fine for something this simple. When I originally wrote this program, I used lists, populated them with true or false values, randomized them, and picked from them over and over again. I am not sure what kind of awful state of mind I was in when I wrote that code (whatever it was, there wasn’t enough caffeine), but I’ve come up with a way more elegant solution – Random.NextDouble(). If the double is less than the probability of rain, it rained. Otherwise, it didn’t. We also need three counters – one for when it rains sometime during the weekend at least once, one for when it rains exactly once and one for when it rains twice.

There is one other case: when it doesn’t rain, but 1,000,000 – the number of times it rained at least once takes care of that. Some operators make determining which case we’re dealing with easy – XOR is true if exactly one of two cases is true, OR is true if at least one case is true, and AND is true if both cases are true.

With that taken care of, all that was left was to write the code in a console application:

static void Main(string[] args)
    double probRainSat = 0.1;
    double probRainSun = 0.3;
    int rainyWeekends = 0;
    int oneDayRainyWeekends = 0;
    int bothDayRainyWeekends = 0;
    Console.WriteLine("Simulating one million weekends...");
    Random rand = new Random(Environment.TickCount);
    // one million trials
    for (int i = 0; i < 1000000; i++)
        if (i % 100000 == 0)
            Console.WriteLine("{0}...", i);
        bool rainedSat = rand.NextDouble() < probRainSat;
        bool rainedSun = rand.NextDouble() < probRainSun;
        if (rainedSun || rainedSat)
        if (rainedSat ^ rainedSun)
        else if(rainedSat && rainedSun) // We can use else if because it can't rain on exactly one and both days.
    Console.WriteLine("Rain on only 1 day: {0}/1,000,000 ({1})", oneDayRainyWeekends, oneDayRainyWeekends / 1000000);
    Console.WriteLine("Rain during the weekend: {0}/1,000,000 ({1})", rainyWeekends, rainyWeekends / 1000000);
    Console.WriteLine("Rain on both days: {0}/1,000,000 ({1})", bothDayRainyWeekends, bothDayRainyWeekends / 1000000);
    Console.WriteLine("Rain on neither day: {0}/1,000,000 ({1})", 1000000 - rainyWeekends, (1000000 - rainyWeekends) / 1000000);
    Console.ReadLine(); // Keep the program from closing.

Now you can run your new creation:

Probability Simulator Results

You just simulated about 19,200 years worth of weekends.

Not bad, huh? If you round, you’re spot on with the calculations. The math predicted almost exactly what the program would do. That, my friends, is pretty cool.