source

​
AutoRandomForest

 AutoRandomForest (config:Optional[Callable[[optuna.trial._trial.Trial],Di
                   ct[str,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoElasticNet

 AutoElasticNet (config:Optional[Callable[[optuna.trial._trial.Trial],Dict
                 [str,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoLasso

 AutoLasso (config:Optional[Callable[[optuna.trial._trial.Trial],Dict[str,
            Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoRidge

 AutoRidge (config:Optional[Callable[[optuna.trial._trial.Trial],Dict[str,
            Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoLinearRegression

 AutoLinearRegression (config:Optional[Callable[[optuna.trial._trial.Trial
                       ],Dict[str,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoCatboost

 AutoCatboost (config:Optional[Callable[[optuna.trial._trial.Trial],Dict[s
               tr,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoXGBoost

 AutoXGBoost (config:Optional[Callable[[optuna.trial._trial.Trial],Dict[st
              r,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
AutoLightGBM

 AutoLightGBM (config:Optional[Callable[[optuna.trial._trial.Trial],Dict[s
               tr,Any]]]=None)
Structure to hold a model and its search space
TypeDefaultDetails
configOptionalNonefunction that takes an optuna trial and produces a configuration
source

​
random_forest_space

 random_forest_space (trial:optuna.trial._trial.Trial)
source

​
elastic_net_space

 elastic_net_space (trial:optuna.trial._trial.Trial)
source

​
lasso_space

 lasso_space (trial:optuna.trial._trial.Trial)
source

​
ridge_space

 ridge_space (trial:optuna.trial._trial.Trial)
source

​
linear_regression_space

 linear_regression_space (trial:optuna.trial._trial.Trial)
source

​
catboost_space

 catboost_space (trial:optuna.trial._trial.Trial)
source

​
xgboost_space

 xgboost_space (trial:optuna.trial._trial.Trial)
source

​
lightgbm_space

 lightgbm_space (trial:optuna.trial._trial.Trial)
source

​
AutoModel

 AutoModel (model:sklearn.base.BaseEstimator,
            config:Callable[[optuna.trial._trial.Trial],Dict[str,Any]])
Structure to hold a model and its search space
TypeDetails
modelBaseEstimatorscikit-learn compatible regressor
configCallablefunction that takes an optuna trial and produces a configuration
source

​
AutoMLForecast

 AutoMLForecast
                 (models:Union[List[__main__.AutoModel],Dict[str,__main__.
                 AutoModel]], freq:Union[int,str],
                 season_length:Optional[int]=None, init_config:Optional[Ca
                 llable[[optuna.trial._trial.Trial],Dict[str,Any]]]=None, 
                 fit_config:Optional[Callable[[optuna.trial._trial.Trial],
                 Dict[str,Any]]]=None, num_threads:int=1)
Hyperparameter optimization helper
TypeDefaultDetails
modelsUnionAuto models to be optimized.
freqUnionpandas’ or polars’ offset alias or integer denoting the frequency of the series.
season_lengthOptionalNoneLength of the seasonal period. This is used for producing the feature space.
Only required if init_config is None.
init_configOptionalNoneFunction that takes an optuna trial and produces a configuration passed to the MLForecast constructor.
fit_configOptionalNoneFunction that takes an optuna trial and produces a configuration passed to the MLForecast fit method.
num_threadsint1Number of threads to use when computing the features.
source

​
AutoMLForecast.fit

 AutoMLForecast.fit
                     (df:Union[pandas.core.frame.DataFrame,polars.datafram
                     e.frame.DataFrame], n_windows:int, h:int,
                     num_samples:int, step_size:Optional[int]=None,
                     input_size:Optional[int]=None,
                     refit:Union[bool,int]=False, loss:Optional[Callable[[
                     Union[pandas.core.frame.DataFrame,polars.dataframe.fr
                     ame.DataFrame],Union[pandas.core.frame.DataFrame,pola
                     rs.dataframe.frame.DataFrame]],float]]=None,
                     id_col:str='unique_id', time_col:str='ds',
                     target_col:str='y',
                     study_kwargs:Optional[Dict[str,Any]]=None,
                     optimize_kwargs:Optional[Dict[str,Any]]=None,
                     fitted:bool=False, prediction_intervals:Optional[mlfo
                     recast.utils.PredictionIntervals]=None)
Carry out the optimization process. Each model is optimized independently and the best one is trained on all data
TypeDefaultDetails
dfUnionSeries data in long format.
n_windowsintNumber of windows to evaluate.
hintForecast horizon.
num_samplesintNumber of trials to run
step_sizeOptionalNoneStep size between each cross validation window. If None it will be equal to h.
input_sizeOptionalNoneMaximum training samples per serie in each window. If None, will use an expanding window.
refitUnionFalseRetrain model for each cross validation window.
If False, the models are trained at the beginning and then used to predict each window.
If positive int, the models are retrained every refit windows.
lossOptionalNoneFunction that takes the validation and train dataframes and produces a float.
If None will use the average SMAPE across series.
id_colstrunique_idColumn that identifies each serie.
time_colstrdsColumn that identifies each timestep, its values can be timestamps or integers.
target_colstryColumn that contains the target.
study_kwargsOptionalNoneKeyword arguments to be passed to the optuna.Study constructor.
optimize_kwargsOptionalNoneKeyword arguments to be passed to the optuna.Study.optimize method.
fittedboolFalseWhether to compute the fitted values when retraining the best model.
prediction_intervalsOptionalNoneConfiguration to calibrate prediction intervals when retraining the best model.
ReturnsAutoMLForecastobject with best models and optimization results
source

​
AutoMLForecast.predict

 AutoMLForecast.predict (h:int, X_df:Union[pandas.core.frame.DataFrame,pol
                         ars.dataframe.frame.DataFrame,NoneType]=None,
                         level:Optional[List[Union[int,float]]]=None)
“Compute forecasts
TypeDefaultDetails
hintNumber of periods to predict.
X_dfUnionNoneDataframe with the future exogenous features. Should have the id column and the time column.
levelOptionalNoneConfidence levels between 0 and 100 for prediction intervals.
ReturnsUnionPredictions for each serie and timestep, with one column per model.
source

​
AutoMLForecast.save

 AutoMLForecast.save (path:Union[str,pathlib.Path])
Save AutoMLForecast objects
TypeDetails
pathUnionDirectory where artifacts will be stored.
ReturnsNone
source

​
AutoMLForecast.forecast_fitted_values

 AutoMLForecast.forecast_fitted_values
                                        (level:Optional[List[Union[int,flo
                                        at]]]=None)
Access in-sample predictions.
TypeDefaultDetails
levelOptionalNoneConfidence levels between 0 and 100 for prediction intervals.
ReturnsUnionDataframe with predictions for the training set
import time

import pandas as pd
from datasetsforecast.m4 import M4, M4Evaluation, M4Info
from sklearn.linear_model import Ridge
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import OneHotEncoder

def train_valid_split(group):
    df, *_ = M4.load(directory='data', group=group)
    df['ds'] = df['ds'].astype('int')
    horizon = M4Info[group].horizon
    valid = df.groupby('unique_id').tail(horizon).copy()
    train = df.drop(valid.index).reset_index(drop=True)
    return train, valid

ridge_pipeline = make_pipeline(
    ColumnTransformer(
        [('encoder', OneHotEncoder(), ['unique_id'])],
        remainder='passthrough',
    ),
    Ridge()
)
auto_ridge = AutoModel(ridge_pipeline, lambda trial: {f'ridge__{k}': v for k, v in ridge_space(trial).items()})

optuna.logging.set_verbosity(optuna.logging.ERROR)
group = 'Weekly'
train, valid = train_valid_split(group)
train['unique_id'] = train['unique_id'].astype('category')
valid['unique_id'] = valid['unique_id'].astype(train['unique_id'].dtype)
info = M4Info[group]
h = info.horizon
season_length = info.seasonality
auto_mlf = AutoMLForecast(
    freq=1,
    season_length=season_length,
    models={
        'lgb': AutoLightGBM(),
        'ridge': auto_ridge,
    },
    fit_config=lambda trial: {'static_features': ['unique_id']},
    num_threads=2,
)
auto_mlf.fit(
    df=train,
    n_windows=2,
    h=h,
    num_samples=2,
    optimize_kwargs={'timeout': 60},
    fitted=True,
    prediction_intervals=PredictionIntervals(n_windows=2, h=h),
)
auto_mlf.predict(h, level=[80])
unique_iddslgblgb-lo-80lgb-hi-80ridgeridge-lo-80ridge-hi-80
0W1218035529.43522435061.83536235997.03508636110.92120235880.44509736341.397307
1W1218135521.76489434973.03561736070.49417136195.17575736051.01381136339.337702
2W1218235537.41726834960.05093936114.78359636107.52885235784.06216936430.995536
3W1218335538.05820634823.64070636252.47570536027.13924835612.63572536441.642771
4W1218435614.61121134627.02373936602.19868336092.85848935389.69097736796.026000
………………………
4662W99229215071.53697814484.61739915658.45655715319.14622114869.41056715768.881875
4663W99229315058.14527814229.68632215886.60423415299.54955514584.26935216014.829758
4664W99229415042.49343414096.38063615988.60623215271.74471214365.34933816178.140086
4665W99229515042.14484614037.05390416047.23578715250.07050414403.42879116096.712216
4666W99229615038.72904413944.82148016132.63660915232.12780014325.05977616139.195824
auto_mlf.forecast_fitted_values(level=[95])
unique_iddsylgblgb-lo-95lgb-hi-95ridgeridge-lo-95ridge-hi-95
0W1151071.061060.584344599.6183551521.5503341076.990151556.5354921597.444810
1W1161073.731072.669242611.7032521533.6352321083.633276563.1786171604.087936
2W1171066.971072.452128611.4861391533.4181181084.724311564.2696521605.178970
3W1181066.171065.837828604.8718381526.8038181080.127197559.6725381600.581856
4W1191064.431065.214681604.2486911526.1806711080.636826560.1821671601.091485
…………………………
361881W99227915738.5415887.66122815721.23719516054.08526115927.91818115723.22276016132.613603
361882W99228015388.1315755.94378915589.51975615922.36782315841.59906415636.90364216046.294485
361883W99228115187.6215432.22470115265.80066815598.64873515584.46223215379.76681115789.157654
361884W99228215172.2715177.04083115010.61679715343.46486415396.24322315191.54780115600.938644
361885W99228315101.0315162.09080314995.66677015328.51483615335.98246515131.28704415540.677887
import polars as pl

train_pl = pl.from_pandas(train.astype({'unique_id': 'str'}))
auto_mlf = AutoMLForecast(
    freq=1,
    season_length=season_length,
    models={'ridge': AutoRidge()},
    num_threads=2,
)
auto_mlf.fit(
    df=train_pl,
    n_windows=2,
    h=h,
    num_samples=2,
    optimize_kwargs={'timeout': 60},
    fitted=True,
    prediction_intervals=PredictionIntervals(n_windows=2, h=h),
)
auto_mlf.predict(h, level=[80])
unique_iddsridgeridge-lo-80ridge-hi-80
stri64f64f64f64
”W1”218035046.09666334046.6952136045.498116
”W1”218134743.26921633325.84797536160.690457
”W1”218234489.59108632591.25455936387.927614
”W1”218334270.76817932076.50772736465.02863
”W1”218434124.02185731352.45412136895.589593
……………
”W99”229214719.45709613983.30858215455.605609
”W99”229314631.55207713928.87433615334.229818
”W99”229414532.90523913642.84011815422.97036
”W99”229514446.06544313665.08866715227.04222
”W99”229614363.04960413654.22005115071.879157
auto_mlf.forecast_fitted_values(level=[95])
unique_iddsyridgeridge-lo-95ridge-hi-95
stri64f64f64f64f64
”W1”141061.961249.326428488.7652492009.887607
”W1”151071.061246.067836485.5066572006.629015
”W1”161073.731254.027897493.4667182014.589076
”W1”171066.971254.475948493.9147692015.037126
”W1”181066.171248.306754487.7455752008.867933
………………
”W99”227915738.5415754.55881215411.96864516097.148979
”W99”228015388.1315655.78086515313.19069815998.371032
”W99”228115187.6215367.49846815024.90830115710.088635
”W99”228215172.2715172.59142314830.00125615515.18159
”W99”228315101.0315141.03288614798.4427215483.623053