Prediction intervals, what are those? They are used to make predictions about data that is not in your sample. So next, I will explain when it is useful to calculate a prediction interval and how to use such an interval and interpret it in a correct way. Let's go back to our broken tea bag and production stop example. We already performed the first three steps of regression analysis. And this was the Minitab output. We calculated our fitted line and found that the results were significant and that the number of production stops is a strong predictor of the defect tea bags. We also performed a residual check and concluded that the residuals satisfied our assumptions. These are the first three steps of regression analysis. We made a fitted line plot, performed the main analysis and performed a residual check. We now move on to the fourth and final step, calculating the prediction interval. Imagine you want to keep the broken teabags per day, under 50, you can look at the fitted line plot and go to 15. This gives you just under 5 stops per day. However, this is an average. So with 5 stops per day, you will have 15 defect bags on average per day. That means you will have many days with less than 15 defects, but also many days, probably about half, with more than 15 defect bags. Hence, we need to define our question differently. We will ask, what is the maximum number of production stops if we want the number of broken tea bags to be under 15 on most of our days? And statistically we will quantify this on most days as on 97.5% of the days. To find the answer to this question, we must calculate the prediction interval. Now pause your video, load your data before you continue. Your data in Minitab should look like this. To calculate the prediction interval, we have to go back to our Regression analysis menu, which we found under Stat. Regression and we were using the Fitted Line Plot. Maybe it's still there, but otherwise you can fill in the Response which is Bags and the Predictor or influence factor Stops. Under Option is where you'll find the prediction interval here. Click it to get your prediction interval and note that Mnitab automatically works with a 95% level. OK and OK. This is your Fitted Line Plot. Let's take a look at the Minitab output. This is the prediction interval that is computed. The prediction interval is the fitted line plus or minus two times the standard deviation of the residuals. So 95% of the measurements should be within this interval. Also, two and a half percent of the measurements will be above the interval and of course, two and a half percent will be below the interval. Also, if you repeat the measurements frequently, 97.5% of the measurements will be below the upper line of the interval. This is the prediction interval plot that Minitab computed for us. Let's take a look at an easy method to read this graph. If you take your mouse over the central graphing area and you press your right mouse button, you'll get this menu and you can go to Crosshairs. This will give you this sort of cross to navigate the graph, and your coordinates of the center of the cross are given on your top left. We go to 15, which is the maximum number of defect bags that we wanted, and we see what the prediction interval tells us. If you go to the right, you see that for 15 bags, we need at most 3 Stops or actually just under 3 Stops. So, if we want the maximum of broken tea Bags to be 15 in 97.5% of the days, we have to have at most 3 Stops per day. This can also be calculated by filling in the formula. If you fill in 3 Stops in the formula, the mean of the broken tea Bags is approximately 12. 12 plus 2 times the standard deviation of 1.5 gives you 50. So the maximum amount of production Stops, will be 3. Let's summarize. The prediction interval can be used to determine the level of influence factor, which ensures a certain CTQ performance. We can use crosshairs, or the formula, to determine the required value of the influence factor.