Index of Models
CREATE TABLE elm_models( model_name text UNIQUE, model_style int, estimated int, log_likelihood double, ll_constants double, ll_zeros double, optimization_options text, num_cases int, chosen_var_idCA text, weight_var_idCO text, time_estimate double, time_dataread double, time_errors double, iterations int, notes text, spec text, last_changed text DEFAULT CURRENT_TIMESTAMP);
model_name text
The name of the model.
model_style int
the style of the model
- 1 = mnl
- 2 = nl
estimated int
whether the model has been estimated
- 0 = no
- 1 = yes
- -1 = yes but did not converge
log_likelihood double
log likelihood at convergence
ll_constants double
log likelihood for the constants only model
ll_zeros double
log likelihood for the zero parameters model
optimization_options text
a semicolon delimited list of optimization option strings
num_cases int
the number of cases used to estimate the model
chosen_var_idCA text
the column name (in the idCA dataset) of the chosen-ness variable
weight_var_idCO text
the column name (in the idCO dataset) of the case weight
time_estimate double
the time used to estimate the model parameters
time_dataread double
the time used to read the data
time_errors double
the time used to calculate the standard error of the estimate
iterations int
the number of iterations
notes text
a '\n' delimited list of model notes
spec text
a '\n' delimited list of model commands specific to the model type
Model Tables
CREATE TABLE model_XXX ( parameter_name text, table_source text, node_code int, parameter_type int, param_value double, param_constraint int, param_stderr double, param_t_stat double, nest_members blob, bind_item text, bind_ratio double);
Creating and populating this table and the associated row of the elm_models table is a valid way to specify a model.
parameter_name text
the parameter name, in variable_name[@alternative_name] format for beta parameters, logsum name for nests
table_source text
the table in which the variable appears, NULL if not applicable
parameter_type int
the parameter type.
- 0 = NONE
- 0x1 = LINEAR
- 0x2 = LINEAR_ALTSPEC
- 0x4 = GROUP_LINEAR
- 0x8 = GROUP_LINEAR_ALTSPEC
- 0xF = BASIC_VAR
- 0x10 = SPREAD_POINT
- 0x60 = SPREAD_NORMAL
- 0x20 = SPREAD_TRIANGLE
- 0x40 = SPREAD_UNIFORM
- 0x80 = SPREAD_LOGNORMAL
- 0x100 = QUANT
- 0x200 = QUANT_ALTSPEC
- 0x400 = GROUP_QUANT
- 0x800 = GROUP_QUANT_ALTSPEC
- 0x1000 = QUANT_SCALE
- 0x2000 = LOGSUM
param_value double
on input, an initial value for the parameter; on output, the estimated value
param_constraint int
a 1 to indicate the parameter is constrained to the initial value, 0 otherwise (default=0)
param_stderr double
the standard error of the initial parameter (assumes that parameter was estimated already)
bind_item text
the parameter name, in variable_name[@alternative_name] format, of the binding parameter, or null if not applicable
bind_ratio double
the multiplier on the free-floating parameter's value used to create this parameter
nest_members text
the node_codes of the nests’s members, as a comma delimited string of integers stored in blob format
node_code int
the node code of the associated alternative or node
param_t_stat double
the t-statistic of the parameter estimate, when compared against zero (one for logsum parameters)
Options
CREATE TABLE options (name text UNIQUE, value text);
name text UNIQUE
The name of the option. Any option names can be set, the program will simply ignore options that are not valid.
value text
the value of the option