OpenBB/website/content/sdk/reference/forecast/tft.md

title, description, keywords

title	description	keywords
tft	This Docusaurus page provides in-depth information on the use of Temporal Fusion Transformer for forecasting purposes, as implemented through OpenBB finance. It includes details on the model's parameters, usage, and output, along with the source code and chart visualization capabilities. The page is of high relevance for developers, data analysts, and professionals dealing with predictive modeling and data analysis tasks.	Temporal Fusion Transformer forecasting OpenBB finance Temporal Fusion Transformer chart Machine Learning Data analysis Predictive modeling Predictive analytics Data modeling

import HeadTitle from '@site/src/components/General/HeadTitle.tsx';

import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem';

Performs Temporal Fusion Transformer forecasting

Source Code: [link]

openbb.forecast.tft(data: Union[pd.Series, pd.DataFrame], target_column: str = "close", n_predict: int = 5, past_covariates: str = None, train_split: float = 0.85, forecast_horizon: int = 5, input_chunk_length: int = 14, output_chunk_length: int = 5, hidden_size: int = 16, lstm_layers: int = 1, num_attention_heads: int = 4, full_attention: bool = False, dropout: float = 0.1, hidden_continuous_size: int = 8, n_epochs: int = 200, batch_size: int = 32, model_save_name: str = "tft_model", force_reset: bool = True, save_checkpoints: bool = True)

---

Parameters

Name	Type	Description	Default	Optional
data (Union[pd.Series, pd.DataFrame])		Input Data	None	True
target_column	Optional[str]:	Target column to forecast. Defaults to "close".	close	True
n_predict (int, optional)		Days to predict. Defaults to 5.	None	True
train_split (float, optional)		Train/val split. Defaults to 0.85.	None	True
past_covariates (str, optional)		Multiple secondary columns to factor in when forecasting. Defaults to None.	None	True
forecast_horizon (int, optional)		Forecast horizon when performing historical forecasting. Defaults to 5.	None	True
input_chunk_length (int, optional)		Number of past time steps that are fed to the forecasting module at prediction time. Defaults to 14.	None	True
output_chunk_length (int, optional)		The length of the forecast of the model. Defaults to 5.	None	True
hidden_size (int, optional)		Hidden state size of the TFT. Defaults to 16.	None	True
lstm_layers (int, optional)		Number of layers for the Long Short Term Memory Encoder and Decoder. Defaults to 16.	None	True
num_attention_headers (int, optional)		Number of attention heads. Defaults to 4.	None	True
full_attention (bool, optional)		Whether to apply a multi-head attention query. Defaults to False>	None	True
dropout (float, optional)		Fraction of neurons affected by dropout. Defaults to 0.1.	None	True
hidden_continuous_size (int, optional)		Default hidden size for processing continuous variables. Defaults to 8.	None	True
n_epochs (int, optional)		Number of epochs to run during training. Defaults to 200.	None	True
batch_size (int, optional)		Number of samples to pass through network during a single epoch. Defaults to 32.	None	True
model_save_name (str, optional)		The name for the model. Defaults to tft_model	None	True
force_reset (bool, optional)		If set to True, any previously-existing model with the same name will be reset (all checkpoints will be discarded). Defaults to True.	None	True
save_checkpoints (bool, optional)		Whether or not to automatically save the untrained model and checkpoints from training. Defaults to True.	None	True

---

Returns

Type	Description
	Adjusted Data series, List of historical fcast values, List of predicted fcast values, Optional[float] - precision, Fit Prob. TFT model object.

---

Display Temporal Fusion Transformer forecast

Source Code: [link]

openbb.forecast.tft_chart(data: Union[pd.Series, pd.DataFrame], target_column: str = "close", dataset_name: str = "", n_predict: int = 5, past_covariates: str = None, train_split: float = 0.85, forecast_horizon: int = 5, input_chunk_length: int = 14, output_chunk_length: int = 5, hidden_size: int = 16, lstm_layers: int = 1, num_attention_heads: int = 4, full_attention: bool = False, dropout: float = 0.1, hidden_continuous_size: int = 8, n_epochs: int = 200, batch_size: int = 32, model_save_name: str = "tft_model", force_reset: bool = True, save_checkpoints: bool = True, export: str = "", residuals: bool = False, forecast_only: bool = False, start_date: Optional[datetime.datetime] = None, end_date: Optional[datetime.datetime] = None, naive: bool = False, export_pred_raw: bool = False, external_axes: Optional[List[axes]] = None)

---

Parameters

Name	Type	Description	Default	Optional
data (Union[pd.Series, pd.DataFrame])		Input Data	None	True
target_column	Optional[str]:	Target column to forecast. Defaults to "close".	close	True
dataset_name str	None	The name of the ticker to be predicted	None	True
n_predict (int, optional)		Days to predict. Defaults to 5.	None	True
train_split (float, optional)		Train/val split. Defaults to 0.85.	None	True
past_covariates (str, optional)		Multiple secondary columns to factor in when forecasting. Defaults to None.	None	True
forecast_horizon (int, optional)		Forecast horizon when performing historical forecasting. Defaults to 5.	None	True
input_chunk_length (int, optional)		Number of past time steps that are fed to the forecasting module at prediction time. Defaults to 14.	None	True
output_chunk_length (int, optional)		The length of the forecast of the model. Defaults to 5.	None	True
hidden_size (int, optional)		Hidden state size of the TFT. Defaults to 16.	None	True
lstm_layers (int, optional)		Number of layers for the Long Short Term Memory Encoder and Decoder. Defaults to 16.	None	True
num_attention_headers (int, optional)		Number of attention heads. Defaults to 4.	None	True
full_attention (bool, optional)		Whether to apply a multi-head attention query. Defaults to False>	None	True
dropout (float, optional)		Fraction of neurons affected by dropout. Defaults to 0.1.	None	True
hidden_continuous_size (int, optional)		Default hidden size for processing continuous variables. Defaults to 8.	None	True
n_epochs (int, optional)		Number of epochs to run during training. Defaults to 200.	None	True
batch_size (int, optional)		Number of samples to pass through network during a single epoch. Defaults to 32.	None	True
model_save_name (str, optional)		The name for the model. Defaults to tft_model	None	True
force_reset (bool, optional)		If set to True, any previously-existing model with the same name will be reset (all checkpoints will be discarded). Defaults to True.	None	True
save_checkpoints (bool, optional)		Whether or not to automatically save the untrained model and checkpoints from training. Defaults to True.	None	True
residuals	bool	Whether to show residuals for the model. Defaults to False.	False	True
forecast_only	bool	Whether to only show dates in the forecasting range. Defaults to False.	False	True
start_date	Optional[datetime]	The starting date to perform analysis, data before this is trimmed. Defaults to None.	None	True
end_date	Optional[datetime]	The ending date to perform analysis, data after this is trimmed. Defaults to None.	None	True
naive	bool	Whether to show the naive baseline. This just assumes the closing price will be the same as the previous day's closing price. Defaults to False.	False	True
external_axes	Optional[List[plt.axes]]	External axes to plot on	None	True

---

Returns

This function does not return anything

---