Notifications
Clear all
Topic starter
The Positional Accuracy Handbook of NSSDA has suggested that the independent data set shuold be three times more accurate than the expected accuracy of the test data. Why it should be three times more? Any mathematical way to support this statement? ?????ÿ
Posted : June 12, 2023 11:57 pm
My off-the-cuff guess is to allow statistical testing at the 99%/3-sigma level.
In practice that standard is not often met, something that the handbook itself acknowledges if I remember correctly.
Posted : June 13, 2023 8:38 am
