Nayland College - Mathematics Home . Year 9 . Year 10 . Level 1 . Level 2 . L3 Statistics . L3 Maths . L3 Calculus . About . Links

# Model Robustness & Re-predicting values

 8 Using iNZight to test Model Robustness (Excellence) The robustness of the iNZight modeling can be tested by removing some recent data, then predicting the data again using iNZight. The estimates can then be compared with the original (and the confidence intervals). Did the modeling reproduce the data well, or not? Discuss If the actual data values are within the 95% prediction interval for the estimates (predictions) then the model is robust. German Visitors to NZ comparing excel with iNZight predictions (& the iNZight csv) EXEMPLARS: Use iNZight to produce a time series graph and add the trend line. Copy this into your report and discuss - Alcohol Consumptn: Data csv,| Word Worksheet | Google Doc class write up - Travel Destination Data csv | Word Write up | Google Doc class work - Travel Purposes: Data csv - Electricity Production: Data csv | (Website)

### Comparing Actual Data to Re predicted Values

How?

Record the actual data values that you are intending to remove.

Then delete out some data values at the end of the time series.

How?

Filter data - Select cases,

Select Row number, proceed (and enter in row numbers)

You will need to check the row numbers before you start

Then proceed.

Redo the predictions (estimates) as normal

Compare the Estimates and the Actual data values (with considerations of the 95% prediction intervals)

Discuss this (in context)

###### ACTUAL
 Aug 3949 Sept 4331 Oct 6446
###### ESTIMATE

The estimates (Aug 5100, Sept 5600, & Oct 7100 visitors) are all greater than the actual data (Aug 5100, Sept 5600, & Oct 7100 visitors from Germany)

Only the October actual data value (6446 visitors) was within the 95% prediction interval for the estimates (6070 to 8190 visitors). The values for August & September were both below the lower limit of the 95% prediction interval.

This indicates that the iNZight Holt-Winters LOWESS technique is not that robust in this situation. The removal of the last three data values caused the trend line to change to increase after 2012, thus producing higher estimates than the actual data, as can be seen in the graph below:

### Example: Tourist numbers to NZ from Germany

Excel Predictions:

iNZight Predictions

 iNZight fitted lower 95% bound upper 95% bound Excel Prediction Nov-12 10213 9292 11133 11001 Dec-12 12067 11044 13090 11791 Jan-13 13977 12860 15094 13084 Feb-13 14837 13633 16042 14301 Mar-13 12643 11357 13929 12887 Apr-13 8505 7141 9869 9063 May-13 5547 4108 6985 6810 Jun-13 4291 2781 5800 5820 Jul-13 4127 2549 5706 6051 Aug-13 4330 2685 5975 6158 Sep-13 5093 3384 6803 6407 Oct-13 7005 5233 8777 7805 Nov-13 10774 8770 12777 11395 Dec-13 12628 10569 14686 12185 Jan-14 14538 12426 16650 13478 Feb-14 15398 13233 17564 14695 Mar-14 13204 10987 15421 13281

(you tube video by Pricilla Allan)