324
C.H. Reick' A. Grünewald . B. Page
1
1975
1980
~
ctS
~ 1985
1990
1994
training data
validation data
out-of-sample data
Fig. 16.2. Data splitting used for crossvalidation.
Fig. 16.3. Prediction of the magnitude of Cirripedia nauplius with a 16x5x2xl
Neural Network. Input are the data from the last eight weeks of both Cirripedia
nauplius and diatoms. Full line: measurements; dashed line: predictions; dotted
line: prediction error. For training the data splitting no. 16 of Fig. 16.2 has been
used.
C.H. Reick' A. Grünewald . B. Page
1
1975
1980
~
ctS
~ 1985
1990
1994
training data
validation data
out-of-sample data
Fig. 16.2. Data splitting used for crossvalidation.
Fig. 16.3. Prediction of the magnitude of Cirripedia nauplius with a 16x5x2xl
Neural Network. Input are the data from the last eight weeks of both Cirripedia
nauplius and diatoms. Full line: measurements; dashed line: predictions; dotted
line: prediction error. For training the data splitting no. 16 of Fig. 16.2 has been
used.