From the course: Advanced Predictive Modeling: Mastering Ensembles and Metamodeling

Unlock the full course today

Join today to access over 22,600 courses taught by industry experts or purchase this course individually.

Measuring model accuracy: Value estimation

Measuring model accuracy: Value estimation - SPSS Tutorial

From the course: Advanced Predictive Modeling: Mastering Ensembles and Metamodeling

Start my 1-month free trial

Measuring model accuracy: Value estimation

- [Instructor] We need an easy and objective way to compare our value estimation models in order to determine if the ensemble that we built represents progress. Are we doing better with the ensemble than we were doing with the component models? Well let's find out. So I've copied over the data and I've added some formulas. We're going to be using mean absolute error. Now, if we go to this first cell we can see that we're subtracting the actual miles per gallon and subtracting from it the neural nets prediction. This is a residual folks, so the easy way to remember it is just simply observed minus expected. Observed was 18 miles per gallon and the expected here is 16.8. Problem is is that sometimes you're going to get a positive number and sometimes you get a negative number. So we do the absolute value. We don't have the mean absolute error yet, but then we just take this whole column and we calculate the average and we're all set. So, the formula is easy to find on the web and it's…

Contents