Extrapolation is the fundamental problem of data analysis. Extrapolation is the prediction out of the range of data to make the model.
If the model is made of the range between from 10kg to 20kg, the prediction for 30kg is extrapolation.
Extrapolation is the prediction for not validated area.
When we use extrapolation, the information of the range of X and the distance of the area of prediction from the range are useful. Dot-plot and Mahalanobis' Distance are tools to study distance. The idea of Unit Space of MT method is used.
For example, if the range is between from 10kg to 20kg, "Prediction for 30kg may be missed, but for 20kg may be good."
The failure of time depended extrapolation is a deep failure of data analysis.
The introduction of the problem is in the page " Failure of Time Depended Extrapolation ".
NEXT Time Depended ExtrapolationTweet