IMultivariateForecastModel<T>

Interfaces ยท AiDotNet.Interfaces

Marks a multivariate time-series model that forecasts several interrelated series at once (e.g. VAR, VARMA). A multivariate forecast is naturally a [horizon x VariableCount] matrix.

Known implementations: VectorAutoRegressionModel<T>

How It Works

For Beginners: a multivariate model forecasts several series together because they influence each other (GDP and unemployment, temperature and energy use). Forecasting N steps produces N rows, one column per series. The unified Predict front routes such models here and returns the result through the common Vector<T> output (the matrix flattened row-major, with the [horizon, variables] shape recorded on the result so it can be reshaped back to a matrix for the caller โ€” no layout knowledge required).

Properties

Property Summary
int VariableCount { get; } The number of interrelated series (variables) the model forecasts.

Methods

Method Summary
Matrix<T> ForecastMultivariate(Matrix<T> history, int horizon) Forecasts horizon steps for all variables, seeded from the multivariate history the model was trained on.