Hello, I'm Professor Brian Bushee. Welcome to week three of my part of the Accounting Analytics course. This week is gonna be our big data week where we're gonna use some of these big data approaches to try to detect earnings management, building on what we looked at last week. By big data, I mean we're gonna use prediction models to try to predict how the financial statements would look if there was no manipulation by the manager. Which means that any deviations from this expected level we do see are likely to be consistent with earnings management or the manipulation of the financial statements. To do this, we're gonna use a lot of regression models. Now, I know a lot of you haven't seen regression before. That's okay. I'm gonna tell you everything you need to know to be able to do these models. I'll give you all the resources you need. And I'll just, right now, give you a sense for what these regressions represent. So basically, a regression is a way to see how one set of variables helps to explain changes in another variable. For instance, if I wanted to know how much extra salary I would earn in my job if I did more education, I could do a regression model and what it would give me is an estimate of how much my salary would go up if I did one more year of education. So, that's the kind of prediction approach we're gonna use to try to detect earnings management this week. We're gonna start by looking at what are called Discretionary Accruals Models. These models try to model the non-cash portion of earnings or accruals, where managers are making estimates on revenues or expenses. We'll first do a video on estimating the models. Then we'll do a video where you apply the models to some case studies. Next, we'll talk about Discretionary Expenditure Models. These models try to model the cash portion of earnings. These are expenditures on things like research and development, or advertising. Again, we'll start with a video on how to estimate the models. Then we'll look at some refinements to the models and apply them to some case studies. Then we'll look at Fraud Prediction Models. These models try to directly predict what types of companies are likely to commit frauds. Then in the last video, we'll look at something called Benford's Law, which is a little bit off the wall. But what it's gonna do is it's gonna look at the frequency with which certain numbers appear. And if certain numbers appear more often or less often than dictated by Benford's Law, it's gonna be an indication that the financial statements were potentially manipulated. So, these models represent the state of the art right now. These are what academics use to try to detect and predict earnings management. A lot of these models are not easy. I'm gonna try to make them very user friendly. Because by learning this models, you'll come out of this week with a very strong tool kit that will help you try to detect financial statements that may have been manipulated by managers. So, let's get to it.