12/5/2023 0 Comments Fitted value plot minitab expressIf the data is not linear, there will be a pattern to the residuals and this is one way of helping you see that. In response to your comment asking more about the line: On your plot, this means that the point labeled 25 had a predicted value of about 525 but the residual was around -200, meaning its actual value was closer to 325. Remember that a residual is $e_i = y_i - \hat_i$ which is the true $y_i$ minus what the regression estimates should be the outcome for point $i$. In this case it's pretty flat, which provides evidence that a linear model is reasonable. For example, if you fit a linear regression on data that looked like $y = x^2$ you'd see a noticeable bowed shape. Basically, it's smoothing over the points to look for certain kinds of patterns in the residuals. The red line is a LOWESS fit to your residuals vs fitted plot.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |