Let us discuss a couple of distributions of random variables with infinite number of possible values that can be useful in practice. The first one is the geometric distribution, which is a direct generalization of the distribution that we discussed before, a number of coin tossing before first head. So in geometric distribution, we toss a coin but now this coin is not fair for all of them. We toss this coin until the first head appear, X is our random variable, which is the number of tossings, including the one with head. X can take any positive integer value. Let us find the distribution of X. Probability that X takes value K equals to the following. It is the probability of sequence like tail, tail, tail, and so on tail, head, and here we have K minus 1 tails, and the last one case position is head. The probability of outcome of this kind can be calculated by a product of the corresponding outcomes in that we have tail at some particular position, or head at some particular position. So if we assume that the probability of head equals to p, then this probability equals to 1 minus p, to the power k minus 1 times p. This part corresponds to this part, and this p corresponds to this part. Here k can be any positive integer number. Another distribution with infinite number of values is very well known Poisson distribution. Poisson distribution is a distribution of random variable that can take any non-negative integer value, from zero up to infinity. It can be used to model a value that is equal to number of, for example, visitors of your shop during some period of time, for example during the day. It is possible that during some period of time you have zero visitors, if you are unlucky and it is a bad day for you, or it is possible that you have one visitor, or 10 visitors, or 100 visitors. You don't have [inaudible] for the number of visitors. So theoretically, it can be any positive number. Another example, assume that you're a call center, and you want to know what is the number of calls that you have to get and during some period of time. Again, you don't know in advance how many people will call you during some period of time. So you're going to use Poisson random variable to model this number. To specify Poisson distribution, we have to pick some real number, Lambda, which is just arbitrary positive real number, and then we get the probability that our random variable takes value k, and this probability equals to the following thing. This is Lambda to the power k over k factorial times exponent of negative Lambda. This formula looks a little bit cryptic. I will not discuss in details how to get this formula as a limit of binomial distribution under some conditions. I will leave it as an exercise if you're really interested in. But what I want to say is that this Poisson distribution is really useful in practice, and several methods of probability modeling use this Poisson distribution. For example, there exists so-called Poisson regression.