Geoscience Reference
In-Depth Information
Upper Delaware River Watershed
Predicted num. of EPT
0 - 5
5 - 10
11 - 15
16 - 20
21 - 25
New York
Pennsylvania
New Jersey
Fig. 4.6 A map showing the predicted number of EPT individuals for watersheds in the Upper
Delaware river basin, using the same model as that noted in Fig.
4.5
4.5 Discussion
The GAM and MLR model results were similar in terms of the selection of predictor
variables, although the GAMS tended to select fewer variables and performed
consistently better in terms of variance explained. This finding may reflect the
relative advantage of GAMs over linear models in that GAMs require fewer
assumptions of data distributions and error structures, assuming only that functions
are additive and components can be smoothed by local fitting to subsets of the data.
Smoothing parameters were automatically selected based on the effective degrees of
