LinAlgError: Incompatible dimensions

[VedAustin]

from kats.models.theta import ThetaModel, ThetaParams
params = ThetaParams(m=7) # Weekly seasonality 
m = ThetaModel(data_ts, params) # data_ts -> daily data points
m.fit()

gives this error:

Optimization failed to converge. Check mle_retvals.

---------------------------------------------------------------------------
LinAlgError                               Traceback (most recent call last)
/var/folders/25/t45ty5ps70l5y9gww_k11zy40000gq/T/ipykernel_44596/687703454.py in <module>
----> 1 m.fit()

~/kats/lib/python3.9/site-packages/kats/models/theta.py in fit(self, **kwargs)
    119         # creating x and intercept variables to fit a straight line
    120         regr = np.vstack([np.arange(self.n), np.ones(self.n)]).T
--> 121         slope, _ = np.linalg.lstsq(regr, deseas_data.value.values)[0]
    122         # pyre-fixme[16]: `ThetaModel` has no attribute `drift`.
    123         self.drift = slope / 2

<__array_function__ internals> in lstsq(*args, **kwargs)

~/kats/lib/python3.9/site-packages/numpy/linalg/linalg.py in lstsq(a, b, rcond)
   2273     m2, n_rhs = b.shape[-2:]
   2274     if m != m2:
-> 2275         raise LinAlgError('Incompatible dimensions')
   2276 
   2277     t, result_t = _commonType(a, b)

LinAlgError: Incompatible dimensions

Ran the same data under Prophet model and it fits and forecasts w/o errors. Any pointers as to the source of the problem?

[iamxiaodong]

looks like the convergence is failed, do you mind sharing your time series data here so that we can take a look at? thanks.

[VedAustin]

sure .. here you go: data_gh.csv

[VedAustin]

model_params = EnsembleParams(
[
#     BaseModelParams("arima", arima.ARIMAParams(p=1, d=1, q=1)),
#     BaseModelParams("sarima", sarima.SARIMAParams(p=2, d=1, q=1,trend='ct',seasonal_order=(1, 0, 1, 12), enforce_invertibility=False, enforce_stationarity=False)),
    BaseModelParams('prophet', prophet.ProphetParams()),
    BaseModelParams('linear', linear_model.LinearModelParams()),
    BaseModelParams('quadratic', quadratic_model.QuadraticModelParams()),
    BaseModelParams('theta', theta.ThetaParams(m=7))
])

KatsEnsembleParam = {
    'models': model_params,
    'aggregation': 'median',
    'seasonality_length': 7,
    'decomposition_method': 'multiplicative'
}

Errors out at 'arima', 'sarima', 'linear' and 'quadratic' but funnily enough not at 'theta'. Example of error message for 'linear': ValueError: endog and exog matrices are different sizes

Data set same as above.