Struct SarimaModel 

pub struct SarimaModel {
    pub order: SarimaOrder,
    pub ar_coeffs: Array1<f64>,
    pub ma_coeffs: Array1<f64>,
    pub sar_coeffs: Array1<f64>,
    pub sma_coeffs: Array1<f64>,
    pub sigma2: f64,
    pub log_likelihood: f64,
    pub intercept: f64,
    pub std_errors: Option<Array1<f64>>,
    pub transform: Option<f64>,
    pub exog_beta: Option<Array1<f64>>,
}

Fields

order: SarimaOrder

ar_coeffs: Array1<f64>

ma_coeffs: Array1<f64>

sar_coeffs: Array1<f64>

sma_coeffs: Array1<f64>

sigma2: f64

log_likelihood: f64

intercept: f64

Mean of the (differenced) series, added back during forecasting. Estimated only when total differencing d + D <= 1 (a mean for d+D == 0, a drift for d+D == 1); otherwise 0.0.

std_errors: Option<Array1<f64>>

Asymptotic standard errors of the estimated coefficients, in the order [ar.., ma.., sar.., sma..]. None if they could not be computed.

transform: Option<f64>

Box-Cox transform parameter applied to the data before fitting (Some(0.0) is a log transform); forecasts are back-transformed. None means no transform.

exog_beta: Option<Array1<f64>>

Implementations

impl SarimaModel

pub fn fit(data: &Array1<f64>, order: SarimaOrder) -> Result<Self>

Primary constructor for fitting SARIMA/ARIMA models (Conditional Sum of Squares estimation).

pub fn fit_with_method( data: &Array1<f64>, order: SarimaOrder, method: EstimationMethod, ) -> Result<Self>

Fits a SARIMA/ARIMA model with an explicit estimation method.

• EstimationMethod::Css — fast Conditional Sum of Squares (default).
• EstimationMethod::Mle — exact Gaussian maximum likelihood via the Kalman filter; slower but statistically more efficient on short series. Falls back to CSS if the likelihood optimization fails.

pub fn fit_transformed( data: &Array1<f64>, order: SarimaOrder, method: EstimationMethod, transform: Option<f64>, ) -> Result<Self>

Fits a model to Box-Cox-transformed data (transform = lambda; Some(0.0) is a log transform). Requires strictly positive input; forecasts produced by the returned model are automatically back-transformed to the original scale.

pub fn residuals(&self, data: &Array1<f64>) -> Array1<f64>

Returns the model’s in-sample one-step innovation residuals (on the differenced, mean-centered scale the ARMA part was fitted on). These feed the residual diagnostics in crate::diagnostics.

pub fn fit_with_exog( y: &Array1<f64>, exog: Option<&Array2<f64>>, order: SarimaOrder, ) -> Result<Self>

Fits a SARIMAX model. If exog is provided, it first removes the linear trend contributed by X and then fits the SARIMA parameters on the regression residuals.

pub fn forecast_with_intervals_exog( &self, data: &Array1<f64>, exog_hist: Option<&Array2<f64>>, exog_future: Option<&Array2<f64>>, steps: usize, ) -> Result<ForecastResult>

Forecasts a SARIMAX model fitted via Self::fit_with_exog.

The SARIMA dynamics were estimated on the regression residuals eta = y - X * beta, so forecasting requires the historical exogenous matrix exog_hist to reconstruct that residual series, then adds the future exogenous contribution X_fut * beta back onto every band.

For a model fitted without exogenous variables this reduces to Self::forecast_with_intervals; exog_hist/exog_future are ignored.

pub fn forecast_with_intervals( &self, data: &Array1<f64>, steps: usize, ) -> ForecastResult

pub fn fit_arima( data: &Array1<f64>, p: usize, d: usize, q: usize, ) -> Result<Self>

Convenience shortcut for non-seasonal ARIMA

pub fn is_pure_arima(&self) -> bool

Returns true if the fitted model is purely non-seasonal

pub fn aic(&self) -> f64

pub fn aicc(&self, n: usize) -> f64

pub fn bic(&self, n: usize) -> f64

pub fn forecast(&self, history: &Array1<f64>, steps: usize) -> Array1<f64>

Point forecast steps periods ahead. If the model was fitted with a Box-Cox transform, the forecast is automatically back-transformed to the original scale.

Trait Implementations

impl Clone for SarimaModel

fn clone(&self) -> SarimaModel

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SarimaModel

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for SarimaModel

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for SarimaModel

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.