96
P. Goethals . A. Dedecker· W. Gabriels . N. De Pauw
6.2.5
Model Assessment
To compare different models, a validation set of 20 instances was submitted as
input to the models and the predicted outputs were then compared to the measured
results. In this way, the amount of 'correctly classified instances' (CCI) was
obtained and could be used to compare the performance of the different modelling
techniques. Ten fold cross validation (Witten and Frank 2000) for training and
validation on the whole dataset was used in some specific cases and is explicitly
mentioned in the text.
Dissolved oxygen
pR
Flow velo city
Inputlayer
(15 neurons)
Riddon layer
(10 neurons)
Output layer
(52 neurons)
Lumbricidae
Lumbriculidae
Tipulidae
Fig. 6.3. Scheme of the applied neural networks for the Zwalm river data.
P. Goethals . A. Dedecker· W. Gabriels . N. De Pauw
6.2.5
Model Assessment
To compare different models, a validation set of 20 instances was submitted as
input to the models and the predicted outputs were then compared to the measured
results. In this way, the amount of 'correctly classified instances' (CCI) was
obtained and could be used to compare the performance of the different modelling
techniques. Ten fold cross validation (Witten and Frank 2000) for training and
validation on the whole dataset was used in some specific cases and is explicitly
mentioned in the text.
Dissolved oxygen
pR
Flow velo city
Inputlayer
(15 neurons)
Riddon layer
(10 neurons)
Output layer
(52 neurons)
Lumbricidae
Lumbriculidae
Tipulidae
Fig. 6.3. Scheme of the applied neural networks for the Zwalm river data.