@spec add_fourier_features(Explorer.DataFrame.t(), String.t(), map()) ::
  Explorer.DataFrame.t()

Adds Fourier feature columns to a DataFrame for seasonality modeling.

Parameters

• df - An Explorer.DataFrame containing the input data.
• ds_column - The name of the date column.
• seasonality_config - A map containing the seasonality configuration.

Returns

An Explorer.DataFrame with additional columns for Fourier terms.

Examples

iex> df = Explorer.DataFrame.new(%{"ds" => [~D[2023-01-01]], "y" => [1.0]})
iex> config = %{yearly: %{enabled: true, fourier_terms: 2}, weekly: %{enabled: false, fourier_terms: 2}}
iex> result = Soothsayer.Seasonality.add_fourier_features(df, "ds", config)
iex> "yearly_sin_1" in result.names
true

@spec build_components(%{yearly: Axon.t(), weekly: Axon.t()}, map()) :: %{
  yearly: Axon.t(),
  weekly: Axon.t()
}

Builds seasonality component layers.

Parameters

• inputs - Map of Axon input nodes from build_inputs/1.
• config - Model configuration map.

Returns

A map with :yearly and :weekly keys containing Axon layers.

@spec build_features([Date.t()], map()) :: %{
  yearly: Nx.Tensor.t(),
  weekly: Nx.Tensor.t()
}

Builds seasonality feature tensors from dates.

Parameters

• dates - List of dates.
• config - Model configuration map with :seasonality key.

Returns

A map with :yearly and :weekly keys containing Nx tensors.

Examples

iex> dates = [~D[2023-01-01], ~D[2023-01-02], ~D[2023-01-03]]
iex> config = %{seasonality: %{yearly: %{enabled: true, fourier_terms: 2}, weekly: %{enabled: true, fourier_terms: 2}}}
iex> result = Soothsayer.Seasonality.build_features(dates, config)
iex> Nx.shape(result.yearly)
{3, 4}

@spec build_inputs(map()) :: %{yearly: Axon.t(), weekly: Axon.t()}

Creates Axon input nodes for seasonality components.

Parameters

• config - Model configuration map with :seasonality key.

Returns

A map with :yearly and :weekly keys containing Axon input nodes.