sample_parameters_uniformly#

pharmpy.modeling.sample_parameters_uniformly(model, parameter_estimates, fraction=0.1, force_posdef_samples=None, n=1, seed=None, scale='normal')[source]#

Sample parameter vectors using uniform sampling

Each parameter value will be randomly sampled from a uniform distribution with the bounds being estimate ± estimate * fraction.

Parameters:

model (Model) – Pharmpy model
parameter_estimates (pd.Series) – Parameter estimates for parameters to use
fraction (float) – Fraction of estimate value to use for distribution bounds
force_posdef_samples (int) – Number of samples to reject before forcing variability parameters to give positive definite covariance matrices.
n (int) – Number of samples
seed (int or rng) – Random number generator or seed
scale (str) – Scale to perform sampling on. Valid options are ‘normal’ and ‘UCP’

Returns:

pd.DataFrame – samples

Example

>>> from pharmpy.modeling import create_rng, sample_parameters_uniformly, load_example_model
>>> from pharmpy.tools import load_example_modelfit_results
>>> model = load_example_model("pheno")
>>> results = load_example_modelfit_results("pheno")
>>> rng = create_rng(23)
>>> pe = results.parameter_estimates
>>> sample_parameters_uniformly(model, pe, n=3, seed=rng)
      PTVCL      PTVV   THETA_3      IVCL       IVV  SIGMA_1_1
0  0.004878  0.908216  0.149441  0.029179  0.025472   0.012947
1  0.004828  1.014444  0.149958  0.028853  0.027653   0.013348
2  0.004347  1.053837  0.165804  0.028465  0.026798   0.013727