Artificial Intelligence Methods: Toward a New Decision Making Tool

Figures & data

Figure 1. Algorithms of the learning machine .²

Figure 2. Data sets.

Figure 3. Example of a decision tree (AD).⁴

Figure 4. Structure of the neural network.

Figure 5. Support vector regression.

Figure 6. Forecast of the price index by the different models.

Table 1. Descriptive statistics of the price index

	Mean	Std	Median	Kurtosis	Skewness	Jarque-Bera
Price	11 828.94	1 474.18	11 445.22	−0.63	0.60	228.5

Table 2. Parameters of the selected models

Model	Parameter
Tree regression	$depth=36$	$leaves=2882$
Forest regression	$boostrap=True$	$n\mathrm{\_}estimators=100$
SVM	$kernel=poly$	$degree=3$
Neural network (LSTM)	$4layers$	$50neurons/layer$
	$epochs=100$	$optimizer=adam$
Holt-Winters	$method=heuristic$

Table 3. Quadratic error model average

Model	$MSE$
Tree Regression	$651641.78$
Forest Regression	$252295.83$
SVR	$13076.54$
RNN	$12\hspace{0.17em}054.14$
Holt-Winters	$4637230.18$