Ensembles notebook

2b86fbd9638f44cf9b4a3513545ebb58

This notebook contains the simple examples of using the ensemble models with ETNA library.

Table of Contents

[1]:
import warnings
warnings.filterwarnings("ignore")

1. Load Dataset

In this notebook we will work with the dataset contains only one segment with monthly wine sales. Working process with the dataset containing more segments will be absolutely the same.

[2]:
import pandas as pd
from etna.datasets import TSDataset
[3]:
original_df = pd.read_csv("data/monthly-australian-wine-sales.csv")
original_df["timestamp"] = pd.to_datetime(original_df["month"])
original_df["target"] = original_df["sales"]
original_df.drop(columns=["month", "sales"], inplace=True)
original_df["segment"] = "main"
original_df.head()
df = TSDataset.to_dataset(original_df)
ts = TSDataset(df=df, freq="MS")
ts.plot()
../_images/tutorials_ensembles_5_0.png

2. Build Pipelines

Given the sales’ history, we want to select the best model(pipeline) to forecast future sales.

[4]:
from etna.pipeline import Pipeline
from etna.models import NaiveModel, SeasonalMovingAverageModel, CatBoostModelMultiSegment
from etna.transforms import LagTransform
from etna.metrics import MAE, MSE, SMAPE, MAPE
HORIZON = 3
N_FOLDS = 5

Let’s build four pipelines using the different models

[5]:
naive_pipeline = Pipeline(model=NaiveModel(lag=12), transforms=[], horizon=HORIZON)
seasonalma_pipeline = Pipeline(
    model=SeasonalMovingAverageModel(window=5, seasonality=12), transforms=[], horizon=HORIZON
)
catboost_pipeline = Pipeline(
    model=CatBoostModelMultiSegment(),
    transforms=[LagTransform(lags=[6, 7, 8, 9, 10, 11, 12], in_column="target")],
    horizon=HORIZON,
)
pipeline_names = ["naive", "moving average", "catboost"]
pipelines = [naive_pipeline, seasonalma_pipeline, catboost_pipeline]

And evaluate their performance on the backtest

[6]:
metrics = []
for pipeline in pipelines:
    metrics.append(
        pipeline.backtest(
            ts=ts, metrics=[MAE(), MSE(), SMAPE(), MAPE()], n_folds=N_FOLDS, aggregate_metrics=True, n_jobs=5
        )[0].iloc[:, 1:]
    )
metrics = pd.concat(metrics)
metrics.index = pipeline_names
metrics
[Parallel(n_jobs=5)]: Using backend MultiprocessingBackend with 5 concurrent workers.
[Parallel(n_jobs=5)]: Done   1 tasks      | elapsed:    6.3s
[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:   12.0s remaining:   17.9s
[Parallel(n_jobs=5)]: Done   3 out of   5 | elapsed:   17.0s remaining:   11.3s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   26.2s remaining:    0.0s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   26.2s finished
[Parallel(n_jobs=5)]: Using backend MultiprocessingBackend with 5 concurrent workers.
[Parallel(n_jobs=5)]: Done   1 tasks      | elapsed:    5.0s
[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    9.7s remaining:   14.6s
[Parallel(n_jobs=5)]: Done   3 out of   5 | elapsed:   16.5s remaining:   11.0s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   28.1s remaining:    0.0s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   28.1s finished
[Parallel(n_jobs=5)]: Using backend MultiprocessingBackend with 5 concurrent workers.
[Parallel(n_jobs=5)]: Done   1 tasks      | elapsed:    5.9s
[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:   11.4s remaining:   17.0s
[Parallel(n_jobs=5)]: Done   3 out of   5 | elapsed:   16.7s remaining:   11.1s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   28.8s remaining:    0.0s
[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:   28.8s finished
[6]:
MAE MSE SMAPE MAPE
naive 2437.466667 1.089199e+07 9.949886 10.222106
moving average 1913.826667 6.113701e+06 7.897570 7.824056
catboost 2271.766726 8.923741e+06 9.376638 10.013138

3. Ensembles

To improve the performance of the individual models, we can try to make ensembles out of them. Our library contains two ensembling methods, which we will try on now.

3.1 VotingEnsemble

VotingEnsemble forecasts future values with weighted averaging of it’s pipelines forecasts.

[7]:
from etna.ensembles import VotingEnsemble

By default, VotingEnsemble uses uniform weights for the pipelines’ forecasts. However, you can specify the weights manually using the weights parameter. The higher weight the more you trust the base model.

Note: The weights are automatically normalized.

[8]:
voting_ensemble = VotingEnsemble(pipelines=pipelines, weights=[1, 9, 4], n_jobs=4)
[9]:
voting_ensamble_metrics = voting_ensemble.backtest(
    ts=ts, metrics=[MAE(), MSE(), SMAPE(), MAPE()], n_folds=N_FOLDS, aggregate_metrics=True, n_jobs=2
)[0].iloc[:, 1:]
voting_ensamble_metrics.index = ["voting ensemble"]
voting_ensamble_metrics
[Parallel(n_jobs=2)]: Using backend MultiprocessingBackend with 2 concurrent workers.
[Parallel(n_jobs=2)]: Done   1 tasks      | elapsed:    6.6s
[Parallel(n_jobs=2)]: Done   2 tasks      | elapsed:   11.7s
[Parallel(n_jobs=2)]: Done   3 out of   5 | elapsed:   11.7s remaining:    7.8s
[Parallel(n_jobs=2)]: Done   5 out of   5 | elapsed:   12.8s remaining:    0.0s
[Parallel(n_jobs=2)]: Done   5 out of   5 | elapsed:   12.8s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
[9]:
MAE MSE SMAPE MAPE
voting ensemble 1972.207943 6.685831e+06 8.172377 8.299714

3.2 StackingEnsemble

StackingEnsemble forecasts future using the metamodel to combine the forecasts of it’s pipelines.

[10]:
from etna.ensembles import StackingEnsemble

By default, StackingEnsemble uses only the pipelines’ forecasts as features for the final_model. However, you can specify the additional features using the features_to_use parameter. The following values are possible: + None - use only the pipelines’ forecasts(default) + List[str] - use the pipelines’ forecasts + features from the list + “all” - use all the available features

Note: It is possible to use only the features available for the base models.

[11]:
stacking_ensemble_unfeatured = StackingEnsemble(pipelines=pipelines, n_folds=10, n_jobs=4)
[12]:
stacking_ensamble_metrics = stacking_ensemble_unfeatured.backtest(
    ts=ts, metrics=[MAE(), MSE(), SMAPE(), MAPE()], n_folds=N_FOLDS, aggregate_metrics=True, n_jobs=2
)[0].iloc[:, 1:]
stacking_ensamble_metrics.index = ["stacking ensemble"]
stacking_ensamble_metrics
[Parallel(n_jobs=2)]: Using backend MultiprocessingBackend with 2 concurrent workers.
[Parallel(n_jobs=2)]: Done   1 tasks      | elapsed:   14.9s
[Parallel(n_jobs=2)]: Done   2 tasks      | elapsed:   22.5s
[Parallel(n_jobs=2)]: Done   3 out of   5 | elapsed:   25.7s remaining:   17.1s
[Parallel(n_jobs=2)]: Done   5 out of   5 | elapsed:   36.4s remaining:    0.0s
[Parallel(n_jobs=2)]: Done   5 out of   5 | elapsed:   36.4s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    1.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    1.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    2.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    3.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    3.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    4.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    5.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    6.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    7.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.2s finished
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.5s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.5s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    1.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    1.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    2.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    3.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    3.9s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    4.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    5.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    6.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    7.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.4s finished
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.8s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.8s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    1.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    1.0s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.4s finished
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.3s finished
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    1.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    1.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    2.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    2.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    3.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    4.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    5.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    5.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    6.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    7.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    7.3s finished
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    8.4s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    8.4s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.8s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.8s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    1.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    1.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    2.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    3.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    4.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    5.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    5.9s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    6.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    7.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.2s finished
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.7s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.7s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.9s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    0.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    0.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    0.4s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    0.5s finished
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    1.2s remaining:    0.0s
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.8s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    1.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    2.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    3.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   5 out of   5 | elapsed:    4.3s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   6 out of   6 | elapsed:    5.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   7 out of   7 | elapsed:    6.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   8 out of   8 | elapsed:    6.9s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   9 out of   9 | elapsed:    7.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.5s remaining:    0.0s
[Parallel(n_jobs=1)]: Done  10 out of  10 | elapsed:    8.5s finished
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.8s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    9.8s finished
/Users/d.a.binin/Library/Caches/pypoetry/virtualenvs/etna-5hbqKzTr-py3.8/lib/python3.8/site-packages/joblib/parallel.py:735: UserWarning: Multiprocessing-backed parallel loops cannot be nested, setting n_jobs=1
  n_jobs = self._backend.configure(n_jobs=self.n_jobs, parallel=self,
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.7s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.7s finished
[Parallel(n_jobs=4)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=4)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=4)]: Done   3 out of   3 | elapsed:    0.1s finished
[12]:
MAE MSE SMAPE MAPE
stacking ensemble 2058.487868 8.182131e+06 8.508705 8.50082

In addition, it is also possible to specify the final_model. You can use any regression model with the sklearn interface for this purpose.

3.3 Results

Finally, let’s take a look at the results of our experiments

[13]:
metrics = pd.concat(
    [
        metrics,
        voting_ensamble_metrics,
        stacking_ensamble_metrics
    ]
)
metrics
[13]:
MAE MSE SMAPE MAPE
naive 2437.466667 1.089199e+07 9.949886 10.222106
moving average 1913.826667 6.113701e+06 7.897570 7.824056
catboost 2271.766726 8.923741e+06 9.376638 10.013138
voting ensemble 1972.207943 6.685831e+06 8.172377 8.299714
stacking ensemble 2058.487868 8.182131e+06 8.508705 8.500820