Re: Local Weather Forecast

Hey JohnDoe2013.

One thing I suggest you add is the time of the fore-cast since it won't be instantaneous. Your idea of breaking up the time is definitely the right approach.

You should take wait all the time to distributions and average over them for each waiting time to get your final waiting time distribution.

As an example if you had say two distributions (0,5) and (0,10) the total would be 15 and averaging over (0,5) gives (5+5)/15 = 10/15 for (0,5) and 5/15 for (5,10) resulting in a probability density function P(0 < T < 5) = 10/15 = 2/3 and P(10 < T < 15) = 1/3.

Re: Local Weather Forecast

Hi Chiro,

I figured it out. The wait intervals are 10, 15, 5, 15, 10 and 5 minutes respectively. So, in hours, we have:

2 intervals of 1/12 hour

2 intervals of 1/6 hour (2 times as likely as the interval of 1/12 hour)

2 intervals of 1/4 hour (3 times as likely as the interval of 1/12 hour)

So I end up with 3 uniform distributions:

f(x) = 6 for $\displaystyle 0 \le x < \frac{1}{12}$

f(x) = 4 for $\displaystyle \frac{1}{12} \le x < \frac{1}{6}$

f(x) = 2 for $\displaystyle \frac{1}{6} \le x < \frac{1}{4}$

which I then combine into 1.