Given that A happened within n instances ...

[Marko Schmarko]

what's the probability that it will happen again within n' instances?

I assume someone must have figured this out. I assume that if one assumes a gaussian distribution, it must be possible to get an estimate.

e.g.:
If I've walked past 4 parking spaces, and there was 1 open spot, what is the likelihood that there will be at least one more space that is open within the next 100? ... making the unlikely assumption that the cards to normally distributed.

Links would suffice for explanation.

[LuckOfTheDraw]

A complete guess:

With x observed occurances in N chances,

estimating the probability of a at least one occurance in M chances would be:

1 - ((N-x)/N)^M

[Marko Schmarko]

bump.

[Marko Schmarko]

Bump!

[Thythe]

I would suggest putting this in the Probability forum, as you will probably get more responses.

[Windchasers]

Note that your original question is not the same as your example.. You changed n from 4 to 100 part way through.

~w

[Siegmund]

windchasers: it looks like the OP intended n and n' to be two distinct numbers. Would have been clearer if there were called something that didnt look quite so similar.

OP: the answer is going to depend - heavily - on what assumptions you make about the distribution. And "gaussian" is not an option for the spacing between discrete items in a series.

One primitive estimate, mentioned above, is to assume that the ratio observed in the first n observations is representative of the population, and that each observation is independent. (As you pointed out, cars in a parking lot won't be -- they will tend to cluster in the more desirable locations.)

Perhaps you could be more specific what type of problem you're envisioning?