def calc_crit_val(params, options):
df = simulate_truncated_data(params, options)
crit_func = rp.get_crit_func(params, options, df)
crit_val = crit_func(params)
return crit_val
def test_generate_random_model():
"""Test if random model specifications can be simulated and processed."""
params, options = generate_random_model()
df = simulate_truncated_data(params, options)
log_like = get_log_like_func(params, options, df)
crit_val = log_like(params)
assert isinstance(crit_val, float)
def test_simulation_and_estimation_with_different_models():
"""Test the evaluation of the criterion function not at the true parameters."""
# Simulate a dataset
params, options = generate_random_model()
df = simulate_truncated_data(params, options)
# Evaluate at different points, ensuring that the simulated dataset still fits.
crit_func = get_crit_func(params, options, df)
params_ = add_noise_to_params(params, options)
params.equals(params_)
crit_func(params)
def test_invariant_results_for_two_estimations():
params, options = generate_random_model()
df = simulate_truncated_data(params, options)
crit_func = get_crit_func(params, options, df)
# First estimation.
crit_val = crit_func(params)
# Second estimation.
crit_val_ = crit_func(params)
assert crit_val == crit_val_
def test_generate_random_model(seed):
"""Test if random model specifications can be simulated and processed."""
np.random.seed(seed)
params, options = generate_random_model()
df = simulate_truncated_data(params, options)
crit_func = get_crit_func(params, options, df)
crit_val = crit_func(params)
assert isinstance(crit_val, float)
def test_invariant_results_for_two_estimations():
num_agents = np.random.randint(5, 100)
constr = {
"simulation_agents": num_agents,
"n_periods": np.random.randint(1, 4),
"n_lagged_choices": np.random.choice(2),
}
# Simulate a dataset.
params, options = generate_random_model(point_constr=constr)
df = simulate_truncated_data(params, options)
crit_func = get_crit_func(params, options, df)
# First estimation.
crit_val = crit_func(params)
# Second estimation.
crit_val_ = crit_func(params)
assert crit_val == crit_val_
def test_simulation_and_estimation_with_different_models():
"""Test the evaluation of the criterion function not at the true parameters."""
# Set constraints.
num_agents = np.random.randint(5, 100)
constr = {
"simulation_agents": num_agents,
"n_periods": np.random.randint(1, 4),
"edu_max": 15,
"edu_start": [7],
"edu_share": [1],
"n_lagged_choices": np.random.choice(2),
}
# Simulate a dataset
params, options = generate_random_model(point_constr=constr)
df = simulate_truncated_data(params, options)
# Evaluate at different points, ensuring that the simulated dataset still fits.
params, options = generate_random_model(point_constr=constr)
crit_func = get_crit_func(params, options, df)
crit_func(params)
def compute_log_likelihood(params, options):
df = simulate_truncated_data(params, options)
log_like = get_log_like_func(params, options, df)
crit_val = log_like(params)
return crit_val