ar_analysis

ar_analysis(
    data,
    order_p=1,
    include_intercept=True,
    training_time_steps=None,
    forecasting_time_steps=0,
    sampler='DEMCz',
    iterations=3000,
    output_length=10000,
    credible_level=0.9,
    seed=12345,
    thinning_interval=-1,
)

Bayesian autoregressive AR(p) time-series analysis.

Fit an AR(order_p) model to the observed series data and return the posterior forecast curve and credible band.

Parameters

Name	Type	Description	Default
data	array_like	Observed time series.	required
order_p	int	Autoregressive order p.	1
include_intercept	bool	Include an intercept term.	True
training_time_steps	int	Number of training time steps. Defaults to the model’s own default (max(30, floor(0.8 * n))), which is invalid for short series – set it explicitly (e.g. 15) when n is small.	None
forecasting_time_steps	int	Extends the forecast horizon past the observed series.	0
sampler	('DEMCz', 'DEMCzs', 'ARWMH', 'NUTS')	MCMC sampler.	"DEMCz"
iterations	int	Number of post-warmup MCMC iterations.	3000
output_length	int	Number of posterior samples used to build the credible band.	10000
credible_level	float	Credible-interval width.	0.90
seed	int	PRNG seed for the sampler.	12345
thinning_interval	int	MCMC thinning interval; -1 keeps the sampler’s own default.	-1

Returns

Name	Type	Description
	dict	The same shape as univariate_analysis: parameters, mode_curve, mean_curve, lower_ci, upper_ci, aic, bic, dic, rmse.

See Also

univariate_analysis, ma_analysis, arima_analysis, arimax_analysis