Optimal Gasoline Price Predictions: Leveraging the ANFIS Regression Model

Table 3

The hyperparameters for the ANFIS model with PSO optimization.

Hyperparameter	Description	Values
n_mf	Number of membership functions for each input feature	[2, 2, 2, 2]
n_outputs	Number of output labels or classes for the ANFIS model	Determined by data
Problem	Specifies the type of problem (classification “C” or continuous regression)	“C” or none
nPop	Number of particles in the PSO algorithm	500
Epochs	Number of iterations the PSO algorithm will run	100
K	Average size of each particle’s group of informants in PSO	3
phi	Coefficient for calculating confidence coefficients in PSO	2.05
vel_fact	Velocity factor for calculating max/min velocities in PSO	0.5
conf_type	Confinement type for particle velocities in PSO	“RB” (random-back)
IntVar	Specifies which variables should be treated as integers	None or “all”
Normalize	Indicates if the search space should be normalized	False
rad	Normalized radius of the hypersphere centered on the best particle	0.1
mu_delta	Allowed movement range for the mean of membership functions	0.2
s_par	Parameters for the standard deviation of membership functions	[0.5, 0.2]
c_par	Range for the exponent of membership functions	[1.0, 3.0]
A_par	Range for the coefficients of the consequent functions	[−1.0, 1.0]