 So far, we've been building decision models with what are known as static variables. We can introduce a little bit of variation as we saw using solver and the scenario situations, the what if in Excel. But we're limited in what we can do by the number of changes we can make and the time it would take for us to evaluate those changes. Let's go back and look at the street vendor problem. Our vendor, like everyone in business I think, wants to maximize profit. And as we saw, he needs to make a decision as to how many papers to buy. The problem is, he doesn't know how many he will sell. So there's some uncertainty there, there's some risk involved. And in order for him to make the best decision, he needs more information and some way to quantify the risk with the decisions he's going to make. Here is the decision model we built in Excel. And we're giving this data up here again, the things that are fixed by the publisher. And we're assuming that those are not going to change, that they are static. The things that could vary are the demand and the quantity he buys. And we don't know what those are. There's some uncertainty in how much the demand is going to be. The purchase quantity, he can decide what that is, but he doesn't know what the outcome of those decisions are going to be, unless he just assumes that these things are not going to change. The yellow areas are the outputs of those decisions. And of course, the thing he's aiming for is the profit. In the Monte Carlo method, we will be able to simulate the outcomes based on changing some variables. And the key is, do you have some information about these variables that would help you define the outcomes? If you think back to statistics, when we looked at the distributions of a variable, we saw that there are a number of different types of distributions. The two types that you need to remember again, that there are the discrete distributions in which things have to take on specific values. And the key there is, in the histograms, there's spaces. And then there's continuous variables in which the histograms do not have spaces. And most commonly, for continuous, we have the normal distribution. And there's the exponential you may have seen. The triangular and the uniform are quite common. And again, those are replicated over in the discreet with a discrete binomial, discrete uniform, and discrete triangular, etc. It turns out that our news boy kept pretty good records. And looking back over the last year, he has this information about the historical sales of his newspapers. They range from a low of 4,000, roughly, up to about 5,100. And the distribution, he plotted histogram, is somewhat like a normal distribution, a little bit skewed to the upper side. But it's possible to use this information and build a model that reflects this uncertainty in the sales and demand of the newspapers.