weekly_seasonality=False,
holidays_prior_scale=0.001,
)
# Add periods from the theory of tides https://en.wikipedia.org/wiki/Theory_of_tides (additive)
# Period is in days
# Fourier order is how many fourier functions can be used, higher is more complex and unstable
# Short
# model.add_seasonality(name='M4', period=6.21/24, fourier_order=1)
# model.add_seasonality(name='M6', period=4.14/24, fourier_order=1)
# model.add_seasonality(name='M6', period=8.17/24, fourier_order=1)
# Semi-diurnal
model.add_seasonality(name='M2', period=12.4206012/24, fourier_order=1)
model.add_seasonality(name='S2', period=12/24, fourier_order=1)
# model.add_seasonality(name='K2', period=12.65834751/24, fourier_order=1)
# diurnal
model.add_seasonality(name='K1', period=23.93447213/24, fourier_order=2)
# model.add_seasonality(name='O1', period=25.81933871/24, fourier_order=2)
# Monthly and higher
model.add_seasonality(name='Mm', period=27.554631896, fourier_order=2)
# model.add_seasonality(name='quarterly', period=91.25, fourier_order=5)
# model.add_seasonality(name='Ssa', period=182.628180208, fourier_order=1)
# model.add_seasonality(name='Sa', period=365.256360417, fourier_order=1)
model.fit(df_trainp)
