The MNLZ (multinomial logit with zonal or quantitative data for alternatives) uses a slightly modified form of the logit formulation, to ensure positive values for the raw parameters to quantitative variables. This formulation was originally proposed by Daly (1982).
Constructor
mnlz(ec as core, model_name as string)
ec
The core object from which to derive this model.
model_name
The name of the model. Using an existing name will automatically load the existing model.
Methods: Setup
qvar (item_name as string, optional param_value as float, optional constraint as int, optional std_err as float, optional bind_var as string, optional bind_ratio as float) → status as int
This method adds a quantitative variable to the model.
item_name
This is a string describing the variable, in standard name format. Use the keyword "CONSTANT" to include an alternative specific constant.
parameter_value
an initial value for the parameter (default=0)
constraint
a 1 to indicate the parameter is constrained to the initial value, 0 otherwise (default=0)
std_err
the standard error of the initial parameter (assumes that parameter was estimated already)
bind_variable_name
where a parameter is constrained equal to another in the same model uses table_name.variable_name[@alternative_name] format of the other parameter default = null string
bind_ratio
the multiplier on the free-floating parameter's value used to create this parameter (default=1)
status
returns OK on success, OOPS on error.
delete_qvar(item_name as string) → status as int
This method removes a quantitative variable from the model.
theta (optional param_value as float, optional constraint as int, optional std_err as float, optional bind_var as string, optional bind_ratio as float) → status as int
This method controls the coefficient to the quatitative term in the systematic utility function.
parameter_value
an initial value for the parameter (default=0)
constraint
a 1 to indicate the parameter is constrained to the initial value, 0 otherwise (default=0)
std_err
the standard error of the initial parameter (assumes that parameter was estimated already)
bind_variable_name
where a parameter is constrained equal to another in the same model uses table_name.variable_name[@alternative_name] format of the other parameter default = null string
bind_ratio
the multiplier on the free-floating parameter's value used to create this parameter (default=1)
status
returns OK on success, OOPS on error.
References
Daly, Andrew (1982) Estimating Choice Models Containing Attraction Variables, Transportation Research Part B, Vol 16, No 1, pp. 5-15.