depp_sfa documentation
Add your content using reStructuredText syntax. See the
reStructuredText
documentation for details.
API Reference
- class depp_sfa.depp_sfa.SFA(y, x, z=None, id_var=None, time_var=None, fun='prod', intercept=True, lamda0=1, method='teJ', form='linear', dummy_indices=None, inference_method='mle', panel_model='bc92', draws=3000, tune=3000, standardize=True)[source]
Bases:
objectStochastic Frontier Analysis (SFA) estimator.
This class supports Cross-sectional models (ALS77, BC95), Panel Data models (BC92), and True Effects models (Greene 2005). It provides dual inference backends: Frequentist (MLE) and Bayesian (MCMC via PyMC).
- Variables:
FUN_PROD – Constant indicating a production frontier.
FUN_COST – Constant indicating a cost frontier.
TE_teJ – Constant for Jondrow et al. (1982) efficiency decomposition.
TE_te – Constant for Battese & Coelli (1988) efficiency decomposition.
TE_teMod – Constant for a modified efficiency decomposition.
- FUN_COST = 'cost'
- FUN_PROD = 'prod'
- TE_te = 'te'
- TE_teJ = 'teJ'
- TE_teMod = 'teMod'
- get_beta()[source]
Get the estimated coefficients for the frontier equation.
- Returns:
Array of estimated beta coefficients.
- Return type:
numpy.ndarray
- get_lambda()[source]
Get the ratio of standard deviations (lambda = sigma_u / sigma_v).
- Returns:
Lambda value.
- Return type:
float
- get_residuals()[source]
Get the model residuals (epsilon = y - X*beta).
- Returns:
Array of residuals.
- Return type:
numpy.ndarray
- get_sigma2()[source]
Get the total variance of the composite error (sigma^2).
- Returns:
Sigma squared value.
- Return type:
float
- get_technical_efficiency()[source]
Get the technical efficiency scores for all observations.
If Bayesian (PyMC) was used, returns the mean of the posterior TE distribution. Otherwise, applies the user-selected frequentist decomposition.
- Returns:
Array of efficiency scores bounded between 0 and 1.
- Return type:
numpy.ndarray