{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Details and Error Analysis Window of DS Software 6

# Details and Error Analysis Window of DS Software 6 - can...

This preview shows page 1. Sign up to view the full content.

Details and Error Analysis Window of DS Software Every time you run analysis with the forecasting module of the DS software, the second result window that it generates is the "Details and Error Analysis" window. That window is generated regardless of what type of analysis is performed: moving average, weighted moving average, exponential smoothing, trend projection, multiplicative seasonality, regression, multiple regression, etc. ..... In that "Details and Error Analysis" Window pay close attention to the column headings for the error columns that the software reports. One is the actual error: Time series value(i.e., actual value) minus the Forecast: (Y-F) Another is the absolute value of error: |Y-F| The third is error squared: (Y-F)^2 The entries in each column represent error made by the forecasting model. In the first column, the actual value is compare to the forecast made for that period and the difference is computed (Y-F). Since over-estimations and under-estimations of the model might cancel each other out, the total in that column "may" be a small value. You
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: can average the entries in that column. But you should realize that b/c the negatives and positives cancelled each other out, this average is not very representatives of the accuracy of the model. Afterall, an error is an error, whether the forecast is higher or lower than the actual value. In the column where absolute values are reported, you ignore the direction of the error and report the absolute value of error. So when you average that column, you call that measure of accuracy: MAD (Mean Absolute Deviation). MAD is described in one of the notes and comments sections of the chapter. In the other column, the error is squared to ignore the direction of error. Of course, by squaring each error term, we end up with big entries in that column b/c the squaring process magnifies the error terms. But again, you can average them as a measure of forecast accuracy which is called MSE. This has been demosntrated in the book and in the PPTs. Dr. J....
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online