def test_rolling_forecast_cv_bad_splits():
endog = y[:100]
cv = RollingForecastCV(initial=90, step=1, h=4)
gen = cv.split(endog)
expected = [
(np.arange(0, 90), np.array([90, 91, 92, 93])),
(np.arange(1, 91), np.array([91, 92, 93, 94])),
(np.arange(2, 92), np.array([92, 93, 94, 95])),
(np.arange(3, 93), np.array([93, 94, 95, 96])),
(np.arange(4, 94), np.array([94, 95, 96, 97])),
(np.arange(5, 95), np.array([95, 96, 97, 98])),
(np.arange(6, 96), np.array([96, 97, 98, 99])),
]
# should be 7
for i, (train, test) in enumerate(gen):
assert_array_equal(train, expected[i][0])
assert_array_equal(test, expected[i][1])
# assert no extra splits
with pytest.raises(StopIteration):
next(gen)
def test_grouped_model_cross_validate():
metrics = ["smape", "mean_squared_error", "mean_absolute_error"]
expected_columns = (
[f"{met}_mean" for met in metrics]
+ [f"{met}_stddev" for met in metrics]
+ ["grouping_key_columns", "key0"]
)
train = data_generator.generate_test_data(1, 2, 765, "2019-01-01")
grouped_model = GroupedPmdarima(
model_template=AutoARIMA(max_order=5, out_of_sample_size=30),
).fit(train.df, train.key_columns, "y", "ds", silence_warnings=True)
cross_validator = RollingForecastCV(h=90, step=120, initial=365)
cv_metrics = grouped_model.cross_validate(train.df, metrics, cross_validator)
assert len(cv_metrics) == 2
assert set(cv_metrics.columns).issubset(set(expected_columns))
silence_warnings=True,
)
# Save to local directory
save_dir = "/tmp/group_pmdarima/pipeline.gpmd"
pipeline_arima.save(save_dir)
# Load from saved model
loaded_model = GroupedPmdarima.load(save_dir)
print("\nPipeline AutoARIMA results:\n", "-" * 40)
get_and_print_model_metrics_params(loaded_model)
print("\nPredictions:\n", "-" * 40)
prediction = loaded_model.predict(
n_periods=30, alpha=0.2, predict_col="predictions", return_conf_int=True
)
print(prediction.to_string())
print("\nCross validation metric results:\n", "-" * 40)
cross_validator = RollingForecastCV(h=30, step=365, initial=730)
cv_results = loaded_model.cross_validate(
df=training_data,
metrics=["mean_squared_error"],
cross_validator=cross_validator,
error_score=np.nan,
verbosity=3,
)
print(cv_results.to_string())
def test_cv_constructor_value_errors():
with pytest.raises(ValueError):
RollingForecastCV(h=-1), # too low horizon
with pytest.raises(ValueError):
RollingForecastCV(step=-1), # too low step
# -*- coding: utf-8 -*-
from pmdarima.compat.pytest import pytest_error_str
from pmdarima.model_selection import RollingForecastCV, \
SlidingWindowForecastCV, check_cv, train_test_split
from pmdarima.datasets import load_wineind
import pytest
import numpy as np
from numpy.testing import assert_array_equal
y = load_wineind()
@pytest.mark.parametrize(
'cv', [
RollingForecastCV(),
RollingForecastCV(h=4),
RollingForecastCV(initial=150, h=10),
RollingForecastCV(initial=12, h=16, step=7),
]
)
def test_rolling_forecast_cv_passing(cv):
# get all splits
splits = list(cv.split(y))
last_train_step = None
for train, test in splits:
assert test.shape[0] == cv.h
assert test[-1] == train[-1] + cv.h
if last_train_step is not None:
assert train[-1] == last_train_step + cv.step
