Gets the coefficient summary df after applying the given filters. Set any of the parameters to `bool` to enable filtering. - Any argumen

Identifies the degenerate columns, and sets ``self.keep_cols`` and ``self.drop_cols``. Parameters ---------- X : `pan

Fits the normalization transform. Parameters ---------- X : `pandas.DataFrame` Training input data. e.g. each col

A function which turns a train-forecast function to a function which forecast each horizon by fitting its own corresponding model with the goal

Generates a function which creates fourier series matrix for a column of an input df :param col_name: str is the column name in the datafr

Generates df to test change points The generated df is simple zigzag shaped over time with noise, with user specified frequency, length and n

Get the ``info_dict`` dictionary for linear models. A series of functions are used in a flow to get all information needed in linear model su

Returns lagged regressor column names and minimal/maximal lag order. The lag order can be used to check potential imputation in the computatio

Returns lagged regressor column names and minimal/maximal lag order. The lag order can be used to check potential imputation in the computatio

Returns a function that evaluates the logistic function at t with the specified growth rate, capacity, floor, and inflection point. f

Returns the expected argument list

Returns pipeline. Implementation may be overridden by subclass if a different pipeline is desired. Uses ``self.estimator``,

Returns regressor column names from the model components. Currently does not implement regressors.

Gets the regularization strength. The regularization strength typically won't affect the result too much, if set in a reasonable range (0.4-0

Parses the adaptive lasso estimator to get change point dates. The functions calls ``adaptive_lasso_cv`` to selected potential seasonality change

Infers the yearly seasonality order for changepoint detection. Parameters ---------- df : `pandas.DataFrame` The input timeseris.

This function applies limits (lower bound and upper bound) to values (could be any value type for which min and max are defined) given in tupl

Loads hierarchical actuals. This dataset contains synthetic data that satisfy hierarchical constraints. Consider the 3-level tree wit

Loads hierarchical forecasts. This dataset contains forecasts for the actuals given by `~greykite.common.data_loader.DataLoader.load_

This function combines anomalies information that may exist in various places and combine them into a consistent `anomaly_df`: a point will be

Takes a dataframe (df) with at least two columns (`offset_col` and `tuple_col`) and offsets all values given in tuple_col using an offset

Makes component plots. Returns ------- figs : `list` [`plotly.graph_objects.Figure` or None] A list of figures fr

This function operates on a given data frame (``df``) which includes time (given in ``time_col``) and metrics (given in ``value_cols``), as well a

Overlay by splitting wrt values of a column (split_col). If some agg metrics (specified by agg_dict) are also desired, we overlay the

Pretty print the dictionary 'params' Copied from sklearn.base._pprint to avoid accessing protected member of module Parameters ---------

Predicts the interval. Parameters ---------- fut_df : `pandas.DataFrame` The dataframe used for prediction.

A method to prepares the data for ``fit``, ``calc_with_param``. Parameters ---------- data : See class attributes. R

Rounds `split_scores` to the specified `decimals` and returns the result as a list. Parameters ---------- split_score

Runs every config and stores the output of the :func:`~greykite.framework.pipeline.pipeline.forecast_pipeline`. This function runs onl

Scales ``std_col`` and ``quantile_summary_col`` of ``df`` by ``sigma_scaler`` in-place.

Rolls a 1d-array in either direction, and applies a fill value on a portion of the array Parameters ---------- arr : array-like, 1-D

A function to simulate future series. If the fitted model supports uncertainty e.g. via ``uncertainty_dict``, errors are incorporated

Gets model summaries. Returns ------- summaries : `list` [`~greykite.algo.common.model_summary.ModelSummary` or None]

Converts an instance of the `ADConfig` class to its json string format.

Removes duplicate values from ``volatility_features_list`` and validates the features against ``valid_features``. Parameters ---------

For a given set of datarfames with same columns in ``df_dict``, it will concat them vertically by using ``join_cols`` as joining columns.

Addition of objects or an object with a number (scalar).

If forecast_one_by_one is activated, 1. replaces the estimator with ``OneByOneEstimator`` in pipeline. 2. Adds one by one est

Flattens the ``estimator_params_list``. The ``estimator_params_list`` is from ``self.__get_estimators_and_params_from_template_configs``,

Generates the autoregressive components for forecasting given the forecast horizon and time frequency. Only if ``forecast_horizon`` i

Generates the lagged regressor components for forecasting given the forecast horizon and time frequency. This applies to ONE lagged re

Gets the default model components from a model template name. Parameters ---------- template : `str` The model te

Gets the estimators and estimator parameters from ``MultistageForecastTemplateConfig`` objects. Parameters ---------- new_con

Returns default feature sets based on training data frequency and size. Parameters ---------- simple_freq: `str`, default Sim

Defines features sets for use in the `extra_pred_cols` parameter to `forecast_silverkite`. Derived from events, seasonality, and trend

Gets the hyperparameter grid from the `ModelComponentsParam` class. Parameters ---------- model_components : `class`

Gets the `ModelComponentsParam` class from model template, and overridden by ``model_components``. The template could be a name string, a `Si

Gets the default model template when "auto" is given. This is called after ``config`` has been filled with the default values and all

Gets the overriden Multistage Forecast configs by ``custom``. Parameters ---------- custom : `dict` [`str`, any]

Gets the name string(s) from model template. The template could be a name string, a `SimpleSilverkiteTemplateOptions` dataclass, or a

Returns requested seasonality fourier series order. Parameters ---------- requested_seasonality : `str` or `bool` or `int` o

Generates holidays dictionary for input to daily_event_df_dict parameter of silverkite model. The main purpose is to provide reasonable defaul

Generates `fs_components_df` parameter for `forecast_silverkite` for modeling seasonality. Parameters ---------- simp

Gets the `ModelComponentsParam` class from a single simple silverkite template name. Parameters ---------- template : `str`

Extracts template class (e.g. `SimpleSilverkiteTemplate`) from the config. Parameters ---------- config : :class:`~greykite.f

Multiplication of objects or an object with a number.

Normalizes the ``changepoint_values`` in `~greykite.algo.forecast.silverkite.SilverkiteForecast.forecast` with the same normalize meth

Uses the values in ``model_components`` to update the corresponding values in ``default_model_components``. If a value is not set in model_co

Removes fourier series terms with perfect or almost perfect collinearity. This function is intended to be used when fitting algorithm is OLS.

Removes interaction terms that include fourier series terms removed in `~greykite.algo.forecast.silverkite.SilverkiteForecast.__remove_fourier

Constructs constraint matrix that requires each parent nodes's value to be the sum of its leaf nodes' values. Returns -------

Constructs leaf projection matrix. Returns ------- leaf_projection_matrix : `numpy.array`, shape (``self.num_leaf_nodes``, ``

Constructs the summing matrix. Returns ------- sum_matrix : `numpy.array`, shape (``self.num_nodes``, ``self.num_leaf_nodes``

Checks if ``coverage`` is consistent with the ``uncertainty_dict`` used to train the ``forecast_silverkite`` model. Sets ``uncertainty_dict``

Decodes the string format single template name from the ``name_dataclass`` dataclass. Parameters ---------- name_dataclass :

Checks that necessary input are provided in ``self.uncertainty_dict`` and ``self.train_df``. To be called after setting ``self.train_df`` in `

Preprocesses the input design matrix, the response vector, l1_index, l2_index and constant columns. Parameters ---------- x :

Preprocesses the input design matrix, the response vector, l1_index, l2_index and constant columns. Parameters ---------- x :

Deserializes a class. Parameters ---------- serialized_dict: `dict` The serialized dictionary. obj_type:

Deserializes a dictionary. Parameters ---------- serialized_dict: `dict` The serialized dictionary. obj_t

Deserializes a list or a tuple. Parameters ---------- serialized_dict: `dict` The serialized dictionary.

Deserializes an ordered dictionary. Parameters ---------- serialized_dict: `dict` The serialized dictionary.

Deserializes the root object. This is the very top level of the nested `serialized_dict`. Thus, it either contains a single .pkl file

Return non-default make_scorer arguments for repr.

Fits the regression problem with a single L2 norm regularization parameter and cross validation over many L1 norm regularization parameters.

Fits the uncertainty model.

Gets the Lasso alpha grid. Parameters ---------- x : `numpy.array` The design matrix. y : `numpy.array`

Gets the candidate holidays from a list of countries. Uses `pypi:holidays-ext`. Duplicates are dropped. Observed holidays are

Gets the estimators for forecast one-by-one. If the given parameters indicate that multiple estimators are need for the forecast one-

Computes the transformed lasso input quantities. Parameters ---------- x : `numpy.array` The design matrix.

Gets the number of rounds of predictions needed. For example, when the prediction is 3 weeks while the lag is 1 week, it will need to

Gets the score for each day in ``event_dates``. The score is defined as the observation on the day minus the baseline, which is the av

Classifies holidays into model independently, model together and do not model according to their scores. Returns -------

Gets the model summary from input. This function is called during initialization. Returns ------- info_dict : `dict`

Fits the seasonality with Fourier series and find the best order according to a criterion. Parameters ---------- df : `pandas

When the scores are computed, calculates the contributions and classifies holidays into: - model independently - mode

Makes a figure that contains 2-3 subplots. The subplots are: - AIC/BIC vs. Fourier series orders. - De-trended and ag

Makes a plot that includes the following two subplots: - Bar chart for holiday effects grouped by holidays ordered by their holiday effec

The prediction function. Parameters ---------- df : `pandas.DataFrame` The input dataframe, covering the predicti

Processes the training data. Including appending ``past_df``, interpolation, etc. Returns ------- This function modifies clas