AutocorrelationExplorer
Show how a series relates to its own past, one lag at a time.
Forecasting a series from its own history only works when the history says something about the present, and a correlogram is where that shows up. A bar reaching out of the confidence band at lag 12 on monthly data is a year long season; a slow decay across every lag is a trend; bars that stay inside the band throughout mean the series is noise and no amount of model tuning will forecast it.
The two functions answer different questions. The autocorrelation counts
every route from one period to another, so a strong lag 1 echoes into lag
2 and lag 3 whether or not those lags carry anything of their own. The
partial autocorrelation removes the shorter routes and shows what each
lag adds by itself, which is why ARIMA orders are read off it: the last
lag standing out of the band on the PACF is a candidate for p, and on
the ACF for q.
The same reading sets window_size for TimeSeriesWindowConverter,
which is otherwise a guess: a window shorter than the last useful lag
throws away the information the correlogram just found.
Parameters
- n_lags : integer, default=
20 - How many lags to draw. A lag is one step back along the series, so 20 asks how today relates to each of the last 20 periods. More lags than the series can answer are trimmed.
- function : string, default=
both - Which correlogram to draw. The ACF counts every route from one period to another, so a strong lag 1 echoes into lag 2. The PACF removes the shorter routes and shows what each lag adds on its own.
- confidence : number, default=
0.95 - The confidence level of the band drawn around the bars. A bar inside the band is not distinguishable from noise.
Methods
get_results(self, exploration_path: str, options: Dict[str, Any]) -> List[DashAI.back.core.artifacts.Artifact]
AutocorrelationExplorerLoad and return the saved figure for the frontend.
Parameters
- exploration_path : str
- Path to the JSON file saved by
save_notebook. - options : Dict[str, Any]
- Rendering options from the frontend (unused).
Returns
- List[Artifact]
- A single element list with the plotly artifact of the saved figure.
launch_exploration(self, dataset: 'DashAIDataset', explorer_info: DashAI.back.dependencies.database.models.Explorer)
AutocorrelationExplorerDraw the correlogram of the selected series.
Parameters
- dataset : DashAIDataset
- The prepared dataset holding the selected columns.
- explorer_info : Explorer
- Explorer record with the column names and optional display name.
Returns
- plotly.graph_objects.Figure
- One correlogram per requested function, drawn as bars per lag inside a confidence band.
save_notebook(self, notebook_info: DashAI.back.dependencies.database.models.Notebook, explorer_info: DashAI.back.dependencies.database.models.Explorer, save_path: 'Path', result: Any) -> str
AutocorrelationExplorerSave the figure to disk (JSON content, .pickle extension).
Parameters
- notebook_info : Notebook
- The notebook database record (unused).
- explorer_info : Explorer
- The explorer record used for filename generation.
- save_path : Path
- Directory where the file will be saved.
- result : Any
- The Plotly figure returned by
launch_exploration.
Returns
- str
- The path of the saved file as a POSIX string.
validate_columns(cls, explorer_info: DashAI.back.dependencies.database.models.Explorer, column_spec: Dict[str, Dict[str, str]]) -> bool
AutocorrelationExplorerCheck the selection is one date column plus one series.
Parameters
- explorer_info : Explorer
- The database record for the explorer instance, including the selected columns.
- column_spec : Dict[str, Dict[str, str]]
- A mapping from column name to a dict with at least
"type"and"dtype".
Returns
- bool
- True if the selection holds exactly one Date column and exactly one numeric column, and the inherited checks also pass.
get_credential(self, name: str)
ConfigObjectResolve a registered credential component by name.
Parameters
- name : str
- Credential component class name (e.g. "HuggingFaceCredential").
Returns
- BaseCredential
- An instance of the requested credential component.
get_metadata(cls) -> Dict[str, Any]
BaseExplorerGet metadata for the explorer, used by the DashAI frontend.
Returns
- Dict[str, Any]
- Dictionary containing display name, description, image preview path, category, icon, color, allowed semantic types, allowed dtypes, and input cardinality constraints.
get_schema(cls) -> dict
ConfigObjectGenerates the component related Json Schema.
Returns
- dict
- Dictionary representing the Json Schema of the component.
prepare_dataset(self, loaded_dataset: 'DashAIDataset', columns: List[Dict[str, Any]]) -> 'DashAIDataset'
BaseExplorerPrepare the dataset by selecting only the columns needed for this exploration.
Parameters
- loaded_dataset : DashAIDataset
- The full dataset loaded from storage.
- columns : List[Dict[str, Any]]
- List of column descriptor dicts, each containing at least
"columnName". Optional keys:"id","valueType","dataType".
Returns
- DashAIDataset
- Dataset restricted to the requested columns.
validate_and_transform(self, raw_data: dict) -> dict
ConfigObjectIt takes the data given by the user to initialize the model and returns it with all the objects that the model needs to work.
Parameters
- raw_data : dict
- A dictionary with the data provided by the user to initialize the model.
Returns
- dict
- A validated dictionary with the necessary objects.
validate_parameters(cls, params: Dict[str, Any]) -> bool
BaseExplorerValidate explorer parameters against the explorer's schema.
Parameters
- params : Dict[str, Any]
- The configuration parameters to validate.
Returns
- BaseExplorerSchema
- The validated and parsed schema instance. Subclasses that override this method may return a bool to indicate pass/fail without returning the model instance.