Welcome to Flow Forecast’s documentation!¶
Flow Forecast is a deep learning for time series forecasting framework written in PyTorch. Flow Forecast makes it easy to train PyTorch Forecast models on a wide variety of datasets.
-
flood_forecast.explain_model_output.
handle_dl_output
(dl, dl_class: str, datetime_start: datetime.datetime, device: str) → Tuple[torch.Tensor, int][source]¶ - Parameters
dl (Union[CSVTestLoader, TemporalTestLoader]) – The test data-loader. Should be passed directly
dl_class (str) – A string that is the name of DL passef from the params file.
datetime_start (datetime) – The start datetime for the forecast
device (str) – Typical device should be either cpu or cuda
- Returns
Returns a tuple containing either a..
- Return type
Tuple[torch.Tensor, int]
-
flood_forecast.explain_model_output.
deep_explain_model_summary_plot
(model, csv_test_loader: flood_forecast.preprocessing.pytorch_loaders.CSVTestLoader, datetime_start: Optional[datetime.datetime] = None) → None[source]¶ Generate feature summary plot for trained deep learning models Args:
model (object): trained model csv_test_loader (CSVTestLoader): test data loader datetime_start (datetime, optional): start date of the test prediction,
Defaults to None, i.e. using model inference parameters.
-
flood_forecast.explain_model_output.
deep_explain_model_heatmap
(model, csv_test_loader: flood_forecast.preprocessing.pytorch_loaders.CSVTestLoader, datetime_start: Optional[datetime.datetime] = None) → None[source]¶ Generate feature heatmap for prediction at a start time Args:
model ([type]): trained model csv_test_loader ([CSVTestLoader]): test data loader datetime_start (Optional[datetime], optional): start date of the test prediction,
Defaults to None, i.e. using model inference parameters.
- Returns:
None