Homework 4

 

(Input Validation for Exponential Smoothing) (40 PTS) 1.  You need to modify the program for Exponential Smoothing to check for two things and give appropriate feedback if they are violated.

a.  Make sure that the weight is numeric.

b.  Make sure that the weight is strictly greater than zero and less than or equal to one.

 

(Mean Square Error) (60 PTS) 2. A more standard measurement of error is called MSE = Mean Square Error.  You want to modify the APEExponentialSmoothing function to measure MSE rather than APE.  Make sure to change the name.

The function for MSE is as follows.

 

 

Now, in order to get this into units that are meaningful, and to convince yourself that the number you have is valid, you must remember that the units of MSE are squared units.  Thus in order to be back in the scale of the original data you need to take the square root to get some sense of the magnitude of an "average" error.  

In order to get more confidence in your calculation you may want to take the square root of your MSE to see that it is more on a scale you'd expect to see.  But you do not want to take the square root within your function.  Make sure your function returns the MSE, then take the square root of that in another cell.