Skip to content

Results

Result objects for VAR post-estimation output.

FEVDResult

Bases: VARResultBase

Result from forecast error variance decomposition.

Attributes:

Name Type Description
idata InferenceData

ArviZ InferenceData with FEVD draws.
horizon int

Number of FEVD horizons.
var_names list[str]

Names of variables.

hdi(prob=0.89)

HDI for FEVD.

Returns:

Type Description
HDIResult

HDIResult whose lower / upper DataFrames mirror the shape and
HDIResult

labels of median().

median()

Posterior median FEVD.

Returns:

Type Description
DataFrame

DataFrame indexed by horizon (integer 0..H) with a
DataFrame

MultiIndex(['response', 'shock']) on columns.

plot()

Plot FEVD.

to_dataframe()

Convert FEVD to DataFrame (passthrough to median()).

ForecastResult

Bases: VARResultBase

Result from VAR forecasting.

Attributes:

Name Type Description
idata InferenceData

ArviZ InferenceData with forecast draws.
steps int

Number of forecast steps.
var_names list[str]

Names of forecasted variables.

hdi(prob=0.89)

HDI for forecast.

median()

Posterior median forecast.

plot()

Plot forecast fan chart.

to_dataframe()

Convert to long-format DataFrame.

HDIResult

Bases: ImpulsoBaseModel

Structured HDI output with separate lower/upper bounds.

Attributes:

Name Type Description
lower DataFrame

DataFrame of lower HDI bounds.
upper DataFrame

DataFrame of upper HDI bounds.
prob float

HDI probability level.

HistoricalDecompositionResult

Bases: VARResultBase

Result from historical decomposition.

Attributes:

Name Type Description
idata InferenceData

ArviZ InferenceData with decomposition draws.
var_names list[str]

Names of variables.

hdi(prob=0.89)

HDI for historical decomposition.

Returns:

Type Description
HDIResult

HDIResult whose lower / upper DataFrames mirror the shape and
HDIResult

labels of median().

median()

Posterior median historical decomposition.

Returns:

Type Description
DataFrame

DataFrame indexed by a DatetimeIndex over the in-sample period
DataFrame

(after lag-trimming and any start / end filter applied at
DataFrame

decomposition time), with a MultiIndex(['response', 'shock'])
DataFrame

on columns.

plot()

Plot historical decomposition.

to_dataframe()

Convert historical decomposition to DataFrame (passthrough to median()).

IRFResult

Bases: VARResultBase

Result from impulse response function computation.

Attributes:

Name Type Description
idata InferenceData

ArviZ InferenceData with IRF draws.
horizon int

Number of IRF horizons.
var_names list[str]

Names of variables.

hdi(prob=0.89)

HDI for IRF.

Returns:

Type Description
HDIResult

HDIResult whose lower / upper DataFrames mirror the shape and
HDIResult

labels of median().

median()

Posterior median IRF.

Returns:

Type Description
DataFrame

DataFrame indexed by horizon (integer 0..H) with a
DataFrame

MultiIndex(['response', 'shock']) on columns.

plot()

Plot impulse response functions.

to_dataframe()

Convert IRF to DataFrame (passthrough to median()).

LagOrderResult

Bases: ImpulsoBaseModel

Result from lag order selection.

Attributes:

Name Type Description
aic int

Optimal lag order by AIC.
bic int

Optimal lag order by BIC.
hq int

Optimal lag order by Hannan-Quinn.
criteria_table DataFrame

DataFrame of all criteria values by lag order.

summary()

Return the full criteria table.

Returns:

Type Description
DataFrame

DataFrame with information criteria for each lag order.

SVForecastResult

Bases: VARResultBase

Density forecast from a univariate SV model.

Attributes:

Name Type Description
idata InferenceData

InferenceData with 'forecast' in posterior_predictive.
series_name str

Name of the forecast series.
steps int

Number of forecast steps.

hdi(prob=0.89)

Highest-density interval for the density forecast.

Parameters:

Name Type Description Default
prob float

Probability mass for the HDI. Default 0.89.

 0.89

Returns:

Type Description
HDIResult

HDIResult with lower/upper DataFrames for each forecast step.

median()

Posterior median of the density forecast.

Returns:

Type Description
DataFrame

DataFrame of median forecasts indexed by step.

plot()

Plot the density forecast with HDI bands.

Returns:

Type Description
Figure

Matplotlib Figure of the density forecast.

to_dataframe()

Density forecast posterior median as a DataFrame.

Returns:

Type Description
DataFrame

DataFrame of median forecasts indexed by step.

VARResultBase

Bases: ImpulsoBaseModel

Base class for VAR post-estimation results.

Subclasses that hold a single named DataArray in idata.posterior_predictive (IRF, FEVD) declare its key via the class-level _PRIMARY_KEY; this drives the shared _guard_no_time_dim check.

Attributes:

Name Type Description
idata InferenceData

ArviZ InferenceData holding the result draws.

hdi(prob=0.89) abstractmethod

Compute highest density interval.

Parameters:

Name Type Description Default
prob float

Probability mass for the HDI. Default 0.89.

 0.89

median() abstractmethod

Compute posterior median of the result.

plot() abstractmethod

Plot the result. Subclasses must implement.

to_dataframe() abstractmethod

Convert result to a tidy DataFrame.

VolatilityResult

Bases: VARResultBase

Result from univariate SV fit — posterior of conditional SD.

Conditional SD is sigma_t = exp(h_t / 2), where h_t is the posterior log-volatility path.

Attributes:

Name Type Description
idata InferenceData

InferenceData with 'h' in posterior.
series_name str

Name of the fitted series.
index DatetimeIndex

DatetimeIndex aligned with the fitted series.

hdi(prob=0.89)

Highest-density interval for the conditional SD path.

median()

Posterior median of the conditional SD path.

plot()

Plot the posterior volatility path with HDI bands.

to_dataframe()

Conditional SD posterior median as a DataFrame.