def test_brr_mean():
brr_mean = ReplicateEstimator("brr", "mean")
brr_mean.estimate(y_brr,
sample_wgt_brr,
rep_wgt_brr,
conservative=True,
remove_nan=True)
brr_var = brr_mean.variance
brr_stderr = pow(brr_var.get("__none__"), 0.5)
assert np.isclose(brr_stderr, 0.1656454, atol=1e-7)
def test_fay_mean():
fay_mean = ReplicateEstimator("brr", "mean", fay_coef=fay_coef)
fay_mean.estimate(y_fay,
sample_wgt_fay,
rep_wgt_fay,
conservative=False,
remove_nan=True)
fay_var = fay_mean.variance
fay_stderr = pow(fay_var.get("__none__"), 0.5)
assert np.isclose(fay_stderr, 0.1655723, atol=1e-6)
def test_jkn_mean():
jkn_mean = ReplicateEstimator("jackknife", "mean")
jkn_mean.estimate(y_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
conservative=False,
remove_nan=True)
jkn_var = jkn_mean.variance
jkn_stderr = pow(jkn_var.get("__none__"), 0.5)
assert np.isclose(jkn_stderr, 0.2320822, atol=1e-7)
def test_fay_total():
fay_total = ReplicateEstimator("brr", "total", fay_coef=fay_coef)
fay_total.estimate(
female_fay,
sample_wgt_fay,
rep_wgt_fay,
conservative=True,
remove_nan=True,
)
fay_var = fay_total.variance
fay_stderr = pow(fay_var.get("__none__"), 0.5)
assert np.isclose(fay_stderr, 1396159, atol=1e-1)
def test_boot_mean():
boot_mean = ReplicateEstimator("bootstrap", "mean")
boot_mean.estimate(
y_boot,
sample_wgt_boot,
rep_wgt_boot,
conservative=True,
remove_nan=True,
)
boot_var = boot_mean.variance
boot_stderr = pow(boot_var.get("__none__"), 0.5)
assert np.isclose(boot_stderr, 6.530090, atol=1e-6)
def test_brr_prop():
brr_prop = ReplicateEstimator("brr", "proportion")
brr_prop.estimate(z_brr,
sample_wgt_brr,
rep_wgt_brr,
conservative=True,
remove_nan=True)
brr_var = brr_prop.variance
brr_stderr_0 = pow(brr_var.get("__none__")[0.0], 0.5)
brr_stderr_1 = pow(brr_var.get("__none__")[1.0], 0.5)
assert np.isclose(brr_stderr_0, 0.0018145, atol=1e-7)
assert np.isclose(brr_stderr_1, 0.0018145, atol=1e-7)
def test_brr_total():
brr_total = ReplicateEstimator("brr", "total")
brr_total.estimate(
female_brr,
sample_wgt_brr,
rep_wgt_brr,
conservative=True,
remove_nan=True,
)
brr_var = brr_total.variance
brr_stderr = pow(brr_var.get("__none__"), 0.5)
assert np.isclose(brr_stderr, 1396159, atol=1e-1)
def test_boot_total():
boot_total = ReplicateEstimator("bootstrap", "total")
boot_total.estimate(
married_boot,
sample_wgt_boot,
rep_wgt_boot,
conservative=True,
remove_nan=True,
)
boot_var = boot_total.variance
boot_stderr = pow(boot_var.get("__none__"), 0.5)
assert np.isclose(boot_stderr, 18938.38, atol=1e-2)
def test_jkn_total():
jkn_total = ReplicateEstimator("jackknife", "total")
jkn_total.estimate(
female_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
conservative=True,
remove_nan=True,
)
jkn_var = jkn_total.variance
jkn_stderr = pow(jkn_var.get("__none__"), 0.5)
assert np.isclose(jkn_stderr, 1958480.0, atol=1e-1)
def test_fay_prop():
fay_prop = ReplicateEstimator("brr", "proportion", fay_coef=fay_coef)
fay_prop.estimate(z_fay,
sample_wgt_fay,
rep_wgt_fay,
conservative=True,
remove_nan=True)
fay_var = fay_prop.variance
fay_stderr_0 = pow(fay_var.get("__none__")[0.0], 0.5)
fay_stderr_1 = pow(fay_var.get("__none__")[1.0], 0.5)
assert np.isclose(fay_stderr_0, 0.0018143, atol=1e-7)
assert np.isclose(fay_stderr_1, 0.0018143, atol=1e-7)
def test_brr_ratio():
brr_ratio = ReplicateEstimator("brr", "ratio")
brr_ratio.estimate(
y_brr,
sample_wgt_brr,
rep_wgt_brr,
x=x_brr,
conservative=True,
remove_nan=True,
)
brr_var = brr_ratio.variance
brr_stderr = pow(brr_var.get("__none__"), 0.5)
assert np.isclose(brr_stderr, 0.0008904, atol=1e-7)
def test_jkn_ratio():
jkn_ratio = ReplicateEstimator("jackknife", "ratio")
jkn_ratio.estimate(
y_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
x=x_jkn,
conservative=True,
remove_nan=True,
)
jkn_var = jkn_ratio.variance
jkn_stderr = pow(jkn_var.get("__none__"), 0.5)
assert np.isclose(jkn_stderr, 0.0012466, atol=1e-7)
def test_boot_ratio():
boot_ratio = ReplicateEstimator("bootstrap", "ratio")
boot_ratio.estimate(
y_boot,
sample_wgt_boot,
rep_wgt_boot,
x=x_boot,
conservative=True,
remove_nan=True,
)
boot_var = boot_ratio.variance
boot_stderr = pow(boot_var.get("__none__"), 0.5)
assert np.isclose(boot_stderr, 0.0005171, atol=1e-7)
def test_fay_ratio():
fay_ratio = ReplicateEstimator("brr", "ratio", fay_coef=fay_coef)
fay_ratio.estimate(
y_fay,
sample_wgt_fay,
rep_wgt_fay,
x=x_fay,
conservative=True,
remove_nan=True,
)
fay_var = fay_ratio.variance
fay_stderr = pow(fay_var.get("__none__"), 0.5)
assert np.isclose(fay_stderr, 0.0008898, atol=1e-7)
def test_boot_prop():
boot_prop = ReplicateEstimator("bootstrap", "proportion")
boot_prop.estimate(
z_boot,
sample_wgt_boot,
rep_wgt_boot,
conservative=True,
remove_nan=True,
)
boot_var = boot_prop.variance
boot_stderr_0 = pow(boot_var.get("__none__")[0.0], 0.5)
boot_stderr_1 = pow(boot_var.get("__none__")[1.0], 0.5)
assert np.isclose(boot_stderr_0, 0.0028165, atol=1e-7)
assert np.isclose(boot_stderr_1, 0.0028165, atol=1e-7)
def test_jkn_prop():
jkn_prop = ReplicateEstimator("jackknife", "proportion")
jkn_prop.estimate(z_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
conservative=True,
remove_nan=True)
jkn_var = jkn_prop.variance
jkn_stderr_1 = pow(jkn_var.get("__none__")[1], 0.5)
jkn_stderr_2 = pow(jkn_var.get("__none__")[2], 0.5)
jkn_stderr_3 = pow(jkn_var.get("__none__")[3], 0.5)
assert np.isclose(jkn_stderr_1, 0.0234118, atol=1e-7)
assert np.isclose(jkn_stderr_2, 0.0178846, atol=1e-7)
assert np.isclose(jkn_stderr_3, 0.0147892, atol=1e-7)
def test_jkn_mean_d():
jkn_mean_d = ReplicateEstimator("jackknife", "mean")
jkn_mean_d.estimate(
y_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
domain=domain_jkn,
conservative=True,
remove_nan=True,
)
jkn_var_d = jkn_mean_d.variance
jkn_stderr_d1 = pow(jkn_var_d.get(1), 0.5)
jkn_stderr_d2 = pow(jkn_var_d.get(2), 0.5)
jkn_stderr_d3 = pow(jkn_var_d.get(3), 0.5)
jkn_stderr_d4 = pow(jkn_var_d.get(4), 0.5)
assert np.isclose(jkn_stderr_d1, 0.5825663, atol=1e-7)
assert np.isclose(jkn_stderr_d2, 0.3493011, atol=1e-7)
assert np.isclose(jkn_stderr_d3, 0.4890297, atol=1e-7)
assert np.isclose(jkn_stderr_d4, 0.4432099, atol=1e-7)
def test_fay_total_d():
fay_total_d = ReplicateEstimator("brr", "total", fay_coef=fay_coef)
fay_total_d.estimate(
female_fay,
sample_wgt_fay,
rep_wgt_fay,
domain=domain_fay,
conservative=True,
remove_nan=True,
)
fay_var_d = fay_total_d.variance
fay_stderr_d1 = pow(fay_var_d.get(1), 0.5)
fay_stderr_d2 = pow(fay_var_d.get(2), 0.5)
fay_stderr_d3 = pow(fay_var_d.get(3), 0.5)
fay_stderr_d4 = pow(fay_var_d.get(4), 0.5)
assert np.isclose(fay_stderr_d1, 295677.2, atol=1e-1)
assert np.isclose(fay_stderr_d2, 512088.8, atol=1e-1)
assert np.isclose(fay_stderr_d3, 885104.4, atol=1e-1)
assert np.isclose(fay_stderr_d4, 903432.8, atol=1e-1)
def test_jkn_total_d():
jkn_total_d = ReplicateEstimator("jackknife", "total")
jkn_total_d.estimate(
female_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
domain=domain_jkn,
conservative=True,
remove_nan=True,
)
jkn_var_d = jkn_total_d.variance
jkn_stderr_d1 = pow(jkn_var_d.get(1), 0.5)
jkn_stderr_d2 = pow(jkn_var_d.get(2), 0.5)
jkn_stderr_d3 = pow(jkn_var_d.get(3), 0.5)
jkn_stderr_d4 = pow(jkn_var_d.get(4), 0.5)
assert np.isclose(jkn_stderr_d1, 414764.8, atol=1e-1)
assert np.isclose(jkn_stderr_d2, 718338.9, atol=1e-1)
assert np.isclose(jkn_stderr_d3, 1241590.0, atol=1e-1)
assert np.isclose(jkn_stderr_d4, 1267301.0, atol=1e-1)
def test_fay_mean_d():
fay_mean_d = ReplicateEstimator("brr", "mean", fay_coef=fay_coef)
fay_mean_d.estimate(
y_fay,
sample_wgt_fay,
rep_wgt_fay,
domain=domain_fay,
conservative=True,
remove_nan=True,
)
fay_var_d = fay_mean_d.variance
fay_stderr_d1 = pow(fay_var_d.get(1), 0.5)
fay_stderr_d2 = pow(fay_var_d.get(2), 0.5)
fay_stderr_d3 = pow(fay_var_d.get(3), 0.5)
fay_stderr_d4 = pow(fay_var_d.get(4), 0.5)
assert np.isclose(fay_stderr_d1, 0.4174973, atol=1e-7)
assert np.isclose(fay_stderr_d2, 0.2501473, atol=1e-7)
assert np.isclose(fay_stderr_d3, 0.3465335, atol=1e-7)
assert np.isclose(fay_stderr_d4, 0.3173425, atol=1e-7)
def test_brr_mean_d():
brr_mean_d = ReplicateEstimator("brr", "mean")
brr_mean_d.estimate(
y_brr,
sample_wgt_brr,
rep_wgt_brr,
x=x_brr,
domain=domain_brr,
conservative=True,
remove_nan=True,
)
brr_var_d = brr_mean_d.variance
brr_stderr_d1 = pow(brr_var_d.get(1), 0.5)
brr_stderr_d2 = pow(brr_var_d.get(2), 0.5)
brr_stderr_d3 = pow(brr_var_d.get(3), 0.5)
brr_stderr_d4 = pow(brr_var_d.get(4), 0.5)
assert np.isclose(brr_stderr_d1, 0.4185291, atol=1e-7)
assert np.isclose(brr_stderr_d2, 0.2506872, atol=1e-7)
assert np.isclose(brr_stderr_d3, 0.3459414, atol=1e-7)
assert np.isclose(brr_stderr_d4, 0.3182729, atol=1e-7)
def test_brr_total_d():
brr_total_d = ReplicateEstimator("brr", "total")
brr_total_d.estimate(
female_brr,
sample_wgt_brr,
rep_wgt_brr,
x=x_brr,
domain=domain_brr,
conservative=True,
remove_nan=True,
)
brr_var_d = brr_total_d.variance
brr_stderr_d1 = pow(brr_var_d.get(1), 0.5)
brr_stderr_d2 = pow(brr_var_d.get(2), 0.5)
brr_stderr_d3 = pow(brr_var_d.get(3), 0.5)
brr_stderr_d4 = pow(brr_var_d.get(4), 0.5)
assert np.isclose(brr_stderr_d1, 295677.2, atol=1e-1)
assert np.isclose(brr_stderr_d2, 512088.9, atol=1e-1)
assert np.isclose(brr_stderr_d3, 885104.3, atol=1e-1)
assert np.isclose(brr_stderr_d4, 903432.9, atol=1e-1)
def test_fay_ratio_d():
fay_ratio_d = ReplicateEstimator("brr", "ratio", fay_coef=fay_coef)
fay_ratio_d.estimate(
y_fay,
sample_wgt_fay,
rep_wgt_fay,
x=x_fay,
domain=domain_fay,
conservative=True,
remove_nan=True,
)
fay_var_d = fay_ratio_d.variance
fay_stderr_d1 = pow(fay_var_d.get(1), 0.5)
fay_stderr_d2 = pow(fay_var_d.get(2), 0.5)
fay_stderr_d3 = pow(fay_var_d.get(3), 0.5)
fay_stderr_d4 = pow(fay_var_d.get(4), 0.5)
assert np.isclose(fay_stderr_d1, 0.0022835, atol=1e-7)
assert np.isclose(fay_stderr_d2, 0.0014000, atol=1e-7)
assert np.isclose(fay_stderr_d3, 0.0018239, atol=1e-7)
assert np.isclose(fay_stderr_d4, 0.0016543, atol=1e-7)
def test_jkn_ratio_d():
jkn_ratio_d = ReplicateEstimator("jackknife", "ratio")
jkn_ratio_d.estimate(
y_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
x=x_jkn,
domain=domain_jkn,
conservative=True,
remove_nan=True,
)
jkn_var_d = jkn_ratio_d.variance
jkn_stderr_d1 = pow(jkn_var_d.get(1), 0.5)
jkn_stderr_d2 = pow(jkn_var_d.get(2), 0.5)
jkn_stderr_d3 = pow(jkn_var_d.get(3), 0.5)
jkn_stderr_d4 = pow(jkn_var_d.get(4), 0.5)
assert np.isclose(jkn_stderr_d1, 0.0031952, atol=1e-7)
assert np.isclose(jkn_stderr_d2, 0.0019580, atol=1e-7)
assert np.isclose(jkn_stderr_d3, 0.0025715, atol=1e-7)
assert np.isclose(jkn_stderr_d4, 0.0023108, atol=1e-7)
def test_brr_ratio_d():
brr_ratio_d = ReplicateEstimator("brr", "ratio")
brr_ratio_d.estimate(
y_brr,
sample_wgt_brr,
rep_wgt_brr,
x=x_brr,
domain=domain_brr,
conservative=True,
remove_nan=True,
)
brr_var_d = brr_ratio_d.variance
brr_stderr_d1 = pow(brr_var_d.get(1), 0.5)
brr_stderr_d2 = pow(brr_var_d.get(2), 0.5)
brr_stderr_d3 = pow(brr_var_d.get(3), 0.5)
brr_stderr_d4 = pow(brr_var_d.get(4), 0.5)
assert np.isclose(brr_stderr_d1, 0.0022865, atol=1e-7)
assert np.isclose(brr_stderr_d2, 0.0014021, atol=1e-7)
assert np.isclose(brr_stderr_d3, 0.0018216, atol=1e-7)
assert np.isclose(brr_stderr_d4, 0.0016590, atol=1e-7)
def test_boot_mean_d():
boot_mean_d = ReplicateEstimator("bootstrap", "mean")
boot_mean_d.estimate(
y_boot,
sample_wgt_boot,
rep_wgt_boot,
domain=domain_boot,
conservative=True,
remove_nan=True,
)
boot_var_d = boot_mean_d.variance
boot_stderr_d1 = pow(boot_var_d.get(1), 0.5)
boot_stderr_d2 = pow(boot_var_d.get(2), 0.5)
boot_stderr_d3 = pow(boot_var_d.get(3), 0.5)
boot_stderr_d4 = pow(boot_var_d.get(4), 0.5)
boot_stderr_d5 = pow(boot_var_d.get(5), 0.5)
assert np.isclose(boot_stderr_d1, 18.47222, atol=1e-5)
assert np.isclose(boot_stderr_d2, 11.90187, atol=1e-5)
assert np.isclose(boot_stderr_d3, 12.12145, atol=1e-5)
assert np.isclose(boot_stderr_d4, 15.77978, atol=1e-5)
assert np.isclose(boot_stderr_d5, 27.63810, atol=1e-5)
def test_boot_total_d():
boot_total_d = ReplicateEstimator("bootstrap", "total")
boot_total_d.estimate(
married_boot,
sample_wgt_boot,
rep_wgt_boot,
domain=domain_boot,
conservative=False,
remove_nan=True,
)
boot_var_d = boot_total_d.variance
boot_stderr_d1 = pow(boot_var_d.get(1), 0.5)
boot_stderr_d2 = pow(boot_var_d.get(2), 0.5)
boot_stderr_d3 = pow(boot_var_d.get(3), 0.5)
boot_stderr_d4 = pow(boot_var_d.get(4), 0.5)
boot_stderr_d5 = pow(boot_var_d.get(5), 0.5)
assert np.isclose(boot_stderr_d1, 11679.59, atol=1e-2)
assert np.isclose(boot_stderr_d2, 22394.10, atol=1e-2)
assert np.isclose(boot_stderr_d3, 23855.78, atol=1e-2)
assert np.isclose(boot_stderr_d4, 20428.32, atol=1e-2)
assert np.isclose(boot_stderr_d5, 14252.84, atol=1e-2)
def test_boot_total_d():
boot_total_d = ReplicateEstimator("bootstrap", "total")
boot_total_d.estimate(
married_boot,
sample_wgt_boot,
rep_wgt_boot,
domain=domain_boot,
conservative=True,
remove_nan=True,
)
boot_var_d = boot_total_d.variance
boot_stderr_d1 = pow(boot_var_d.get(1), 0.5)
boot_stderr_d2 = pow(boot_var_d.get(2), 0.5)
boot_stderr_d3 = pow(boot_var_d.get(3), 0.5)
boot_stderr_d4 = pow(boot_var_d.get(4), 0.5)
boot_stderr_d5 = pow(boot_var_d.get(5), 0.5)
assert np.isclose(boot_stderr_d1, 11679.59, atol=1e-2)
assert np.isclose(boot_stderr_d2, 22400.91, atol=1e-2)
assert np.isclose(boot_stderr_d3, 23877.00, atol=1e-2)
assert np.isclose(boot_stderr_d4, 20471.85, atol=1e-2)
assert np.isclose(boot_stderr_d5, 14273.63, atol=1e-2)
def test_boot_ratio_d():
boot_ratio_d = ReplicateEstimator("bootstrap", "ratio")
boot_ratio_d.estimate(
y_boot,
sample_wgt_boot,
rep_wgt_boot,
x=x_boot,
domain=domain_boot,
conservative=True,
remove_nan=True,
)
boot_var_d = boot_ratio_d.variance
boot_stderr_d1 = pow(boot_var_d.get(1), 0.5)
boot_stderr_d2 = pow(boot_var_d.get(2), 0.5)
boot_stderr_d3 = pow(boot_var_d.get(3), 0.5)
boot_stderr_d4 = pow(boot_var_d.get(4), 0.5)
boot_stderr_d5 = pow(boot_var_d.get(5), 0.5)
assert np.isclose(boot_stderr_d1, 0.0015745, atol=1e-7)
assert np.isclose(boot_stderr_d2, 0.0009610, atol=1e-7)
assert np.isclose(boot_stderr_d3, 0.0009393, atol=1e-7)
assert np.isclose(boot_stderr_d4, 0.0012156, atol=1e-7)
assert np.isclose(boot_stderr_d5, 0.0021456, atol=1e-7)
domain=domain_jkn,
conservative=True,
remove_nan=True,
)
jkn_var_d = jkn_mean_d.variance
jkn_stderr_d1 = pow(jkn_var_d.get(1), 0.5)
jkn_stderr_d2 = pow(jkn_var_d.get(2), 0.5)
jkn_stderr_d3 = pow(jkn_var_d.get(3), 0.5)
jkn_stderr_d4 = pow(jkn_var_d.get(4), 0.5)
assert np.isclose(jkn_stderr_d1, 0.5825663, atol=1e-7)
assert np.isclose(jkn_stderr_d2, 0.3493011, atol=1e-7)
assert np.isclose(jkn_stderr_d3, 0.4890297, atol=1e-7)
assert np.isclose(jkn_stderr_d4, 0.4432099, atol=1e-7)
jkn_total = jkn_mean = ReplicateEstimator("jackknife", "total")
def test_jkn_total():
jkn_total = ReplicateEstimator("jackknife", "total")
jkn_total.estimate(
female_jkn,
sample_wgt_jkn,
rep_wgt_jkn,
conservative=True,
remove_nan=True,
)
jkn_var = jkn_total.variance
jkn_stderr = pow(jkn_var.get("__none__"), 0.5)
assert np.isclose(jkn_stderr, 1958480.0, atol=1e-1)