Table 1.
Dietary composition of the basal diet used for the study cats.
Table 2.
Dietary composition of 15% remaining diets used in palatability trials in the study cats.
Figure 1.
Environmental (e.g. ambient temperature and daylight length) and physiological (e.g. body weight and food intake from the study cats) parameters during the course of the study.
Bodyweight (in kg) is expressed as median (black circles) and range (error bars) kg); daylight length (in min/day) is expressed as mean (open circles) and standard error (error bars) for each month; daily temperature (in Celsius) is expressed as mean (solid line and black circles) and minimum and maximum (dashed lines and open circles above and below the mean line); and food intake (in grams/day) is expressed as mean (white columns) and standard deviation (error bars) intake. Daylight length, temperature and food intake showed a seasonal pattern (P<0.001 for all). In contrast no seasonal pattern was observed for body weight, which increased steadily during the course of the study.
Figure 2.
Determination of the best multi-layers perceptron (MLP) architecture (i.e optimal number of hidden neurons).
The curves show mean squared errors in function of the number of hidden neurons for the training (MSEt, black triangles) and the validation phase (MSEv, open circles) of the modelling process. MSEt and MSEv were averaged over the 1000 simulations carried at each level of hidden neurons, with the following predictive variables: year, temperature, daylight length, temperature x daylight length, body weight. As the number of neurons increased, training performance improved (decreasing MSEt), whilst validation performance firstly improved (decreasing MSEv) but then worsened (MSEv increasing) at 8 hidden neurons. Thus, the optimal state was reached at 7 neurons.
Figure 3.
Measured and predicted average food intake, over the 4-year study by using artificial neural network (ANN) so-called multi-layers perceptron (MLP).
Open circles and dashed lines represent the measured intake. The thick solid line represents the MLP average, the solid thin line represents the best MLP, Finally, the area shaded in grey represents the 95% CI for the MLP prediction. As the number of neurons was increased, the training performance improved (decreasing MSEt). The best input variables to predict food intake were year, temperature, daylight length, temperature×daylight length and body weight.
Table 3.
Comparison of input variables combinations to predict food intake using artificial neural networks.