def test_automatic_changepoints_manning():
df = pd.read_csv("examples/example_wp_log_peyton_manning.csv")
m = PMProphet(df, auto_changepoints=True, growth=True, intercept=True, name="model")
m.add_seasonality(seasonality=365, fourier_order=3)
m.fit(method=Sampler.METROPOLIS)
m.predict(60, alpha=0.2, include_history=True, plot=True)
m.plot_components(intercept=False)
def pm_prophet(train,test, column):
# Fit both growth and intercept
train = train.reset_index()
train['date'] = pd.to_datetime(train['date'])
train.columns = ['ds', 'y']
start = time.time()
m = PMProphet(train, growth=True, intercept=True, n_change_points=2, name='model')
# Add monthly seasonality (order: 3)
m.add_seasonality(seasonality=30, order=3)
# Add weekly seasonality (order: 3)
m.add_seasonality(seasonality=7, order=3)
# Fit the model (using NUTS, 1e+4 samples and no MAP init)
m.fit(
draws=10**4,
method='NUTS',
map_initialization=False,
)
ddf = m.predict(7, alpha=0.4, include_history=True, plot=True)
m.plot_components(
intercept=False,
)
pred = inverse_transform(ddf[-7:], 'y_hat')
rmse_pm = sqrt(mean_squared_error(test, pred))
end = time.time()
time_fb = end - start
print('\n The total RMSE using FB prophet : %.3f' % rmse_pm)
result.loc[len(result)] = [column, 'pm_prophet', rmse_pm, time_pm]
#test_plt((20,10), test_df, predictions, column, 'pmprophet')
result_plots['pm_prophet'] = pred
result.loc[len(result)] = [' ', ' ', ' ', ' ']
def test_manning():
df = pd.read_csv("examples/example_wp_log_peyton_manning.csv")
df = df.head(180)
m = PMProphet(df, growth=True, intercept=True, name="model")
m.add_seasonality(seasonality=30, fourier_order=3)
m.fit(method=Sampler.METROPOLIS, draws=2000)
m.predict(60, alpha=0.2, include_history=True, plot=True)
m.plot_components(intercept=False)
def test_manning_reduced_six_months():
df = pd.read_csv("examples/example_wp_log_peyton_manning.csv")
df = df.head(180)
m = PMProphet(df, auto_changepoints=True, growth=True, intercept=True, name="model")
m.add_seasonality(seasonality=7, fourier_order=3)
m.add_seasonality(seasonality=30, fourier_order=3)
m.fit()
m.predict(60, alpha=0.2, include_history=True, plot=True)
m.plot_components(intercept=False)
def test_seasonality_shape():
"""
Verify that that the size of the fourier timeseries is correct
"""
df = pd.read_csv("examples/example_wp_log_peyton_manning.csv")
m = PMProphet(df, auto_changepoints=True, growth=True, intercept=True, name="model")
m.add_seasonality(seasonality=3, fourier_order=3, mode=Seasonality.ADDITIVE)
m.add_seasonality(seasonality=30, fourier_order=3, mode=Seasonality.ADDITIVE)
m.fit(method=Sampler.METROPOLIS, draws=2000)
m.predict(60, alpha=0.2, include_history=True, plot=False)
assert m.trace['seasonality_model'].shape[1] == 12 # (3 + 3) * 2 (sin and cos)
def test_multiplicative_seasonality():
z = np.sin(np.linspace(0, 200, 200)) * np.linspace(0, 200, 200)
df = pd.DataFrame()
df["ds"] = pd.date_range(start="2018-01-01", periods=200)
df["y"] = z
m = PMProphet(df, auto_changepoints=False, growth=True, intercept=False, name="model")
with m.model:
m.priors['growth'] = pm.Constant('growth_model', 1)
m.add_seasonality(seasonality=3.14 * 2, fourier_order=3, mode=Seasonality.MULTIPLICATIVE)
m.fit()
m.predict(60, alpha=0.2, include_history=True, plot=True)
m.plot_components(intercept=False)
def test_multiplicative_seasonality():
df = pd.read_csv("examples/example_wp_log_peyton_manning.csv")
m = PMProphet(df,
auto_changepoints=True,
growth=True,
intercept=True,
name="model")
m.add_seasonality(seasonality=365,
fourier_order=3,
mode=Seasonality.MULTIPLICATIVE)
m.fit(method=Sampler.METROPOLIS, draws=2000)
m.predict(60, alpha=0.2, include_history=True, plot=True)
m.plot_components(intercept=False)