trend_model

tfts.layers.nbeats_layer.trend_model(theta: Tensor, t_b: Tensor, t_f: Tensor, polynomial_size: int) → Tuple[Tensor, Tensor]

Generate the trend backcast and forecast using polynomial basis functions.

Parameters:

• theta (tf.Tensor) – Basis expansion coefficients

• t_b (tf.Tensor) – Input time index for backcast

• t_f (tf.Tensor) – Output time index for forecast

• polynomial_size (int) – Number of polynomial terms

Returns:

A tuple of (backcast, forecast) tensors

Return type:

Tuple[tf.Tensor, tf.Tensor]