In many cases, only the time series at the lowest level of the hierarchies (bottom time series) are available. HierarchicalForecast has tools to create time series for all hierarchies and also allows you to calculate prediction intervals for all hierarchies. In this notebook we will see how to do it.

!pip install hierarchicalforecast statsforecast

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

# compute base forecast no coherent
from statsforecast.models import ETS, Naive
from statsforecast.core import StatsForecast

#obtain hierarchical reconciliation methods and evaluation
from hierarchicalforecast.methods import BottomUp, MinTrace
from hierarchicalforecast.utils import aggregate, HierarchicalPlot
from hierarchicalforecast.core import HierarchicalReconciliation
from hierarchicalforecast.evaluation import HierarchicalEvaluation

/Users/fedex/miniconda3/envs/hierarchicalforecast/lib/python3.10/site-packages/statsforecast/core.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from tqdm.autonotebook import tqdm

​
Aggregate bottom time series

In this example we will use the Tourism dataset from the Forecasting: Principles and Practice book. The dataset only contains the time series at the lowest level, so we need to create the time series for all hierarchies.

Y_df = pd.read_csv('https://raw.githubusercontent.com/Nixtla/transfer-learning-time-series/main/datasets/tourism.csv')
Y_df = Y_df.rename({'Trips': 'y', 'Quarter': 'ds'}, axis=1)
Y_df.insert(0, 'Country', 'Australia')
Y_df = Y_df[['Country', 'Region', 'State', 'Purpose', 'ds', 'y']]
Y_df['ds'] = Y_df['ds'].str.replace(r'(\d+) (Q\d)', r'\1-\2', regex=True)
Y_df['ds'] = pd.to_datetime(Y_df['ds'])
Y_df.head()
CountryRegionStatePurposedsy
0AustraliaAdelaideSouth AustraliaBusiness1998-01-01135.077690
1AustraliaAdelaideSouth AustraliaBusiness1998-04-01109.987316
2AustraliaAdelaideSouth AustraliaBusiness1998-07-01166.034687
3AustraliaAdelaideSouth AustraliaBusiness1998-10-01127.160464
4AustraliaAdelaideSouth AustraliaBusiness1999-01-01137.448533

The dataset can be grouped in the following non-strictly hierarchical structure.

spec = [
    ['Country'],
    ['Country', 'State'], 
    ['Country', 'Purpose'], 
    ['Country', 'State', 'Region'], 
    ['Country', 'State', 'Purpose'], 
    ['Country', 'State', 'Region', 'Purpose']
]

Using the aggregate function from HierarchicalForecast we can generate: 1. Y_df: the hierarchical structured series y[a,b]τ\mathbf{y}_{[a,b]\tau} 2. S_df: the aggregation constraings dataframe with S[a,b]S_{[a,b]} 3. tags: a list with the ‘unique_ids’ conforming each aggregation level.

Y_df, S_df, tags = aggregate(df=Y_df, spec=spec)
Y_df = Y_df.reset_index()

/Users/fedex/miniconda3/envs/hierarchicalforecast/lib/python3.10/site-packages/sklearn/preprocessing/_encoders.py:828: FutureWarning: `sparse` was renamed to `sparse_output` in version 1.2 and will be removed in 1.4. `sparse_output` is ignored unless you leave `sparse` to its default value.
  warnings.warn(

Y_df.head()
unique_iddsy
0Australia1998-01-0123182.197269
1Australia1998-04-0120323.380067
2Australia1998-07-0119826.640511
3Australia1998-10-0120830.129891
4Australia1999-01-0122087.353380
S_df.iloc[:5, :5]
Australia/ACT/Canberra/BusinessAustralia/ACT/Canberra/HolidayAustralia/ACT/Canberra/OtherAustralia/ACT/Canberra/VisitingAustralia/New South Wales/Blue Mountains/Business
Australia1.01.01.01.01.0
Australia/ACT1.01.01.01.00.0
Australia/New South Wales0.00.00.00.01.0
Australia/Northern Territory0.00.00.00.00.0
Australia/Queensland0.00.00.00.00.0
tags['Country/Purpose']

array(['Australia/Business', 'Australia/Holiday', 'Australia/Other',
       'Australia/Visiting'], dtype=object)

We can visualize the S_df dataframe and Y_df using the HierarchicalPlot class as follows.

hplot = HierarchicalPlot(S=S_df, tags=tags)

hplot.plot_summing_matrix()


hplot.plot_hierarchically_linked_series(
    bottom_series='Australia/ACT/Canberra/Holiday',
    Y_df=Y_df.set_index('unique_id')
)

​
Split Train/Test sets

We use the final two years (8 quarters) as test set.

Y_test_df = Y_df.groupby('unique_id').tail(8)
Y_train_df = Y_df.drop(Y_test_df.index)

Y_test_df = Y_test_df.set_index('unique_id')
Y_train_df = Y_train_df.set_index('unique_id')

Y_train_df.groupby('unique_id').size()

unique_id
Australia                                                72
Australia/ACT                                            72
Australia/ACT/Business                                   72
Australia/ACT/Canberra                                   72
Australia/ACT/Canberra/Business                          72
                                                         ..
Australia/Western Australia/Experience Perth/Other       72
Australia/Western Australia/Experience Perth/Visiting    72
Australia/Western Australia/Holiday                      72
Australia/Western Australia/Other                        72
Australia/Western Australia/Visiting                     72
Length: 425, dtype: int64

​
Computing Base Forecasts

The following cell computes the base forecasts for each time series in Y_df using the AutoETS and model. Observe that Y_hat_df contains the forecasts but they are not coherent. Since we are computing prediction intervals using bootstrapping, we only need the fitted values of the models.

fcst = StatsForecast(df=Y_train_df, 
                     models=[ETS(season_length=4, model='ZAA')],
                     freq='QS', n_jobs=-1)
Y_hat_df = fcst.forecast(h=8, fitted=True)
Y_fitted_df = fcst.forecast_fitted_values()

/Users/fedex/miniconda3/envs/hierarchicalforecast/lib/python3.10/site-packages/statsforecast/models.py:526: FutureWarning: `ETS` will be deprecated in future versions of `StatsForecast`. Please use `AutoETS` instead.
  ETS._warn()

​
Reconcile Base Forecasts

The following cell makes the previous forecasts coherent using the HierarchicalReconciliation class. Since the hierarchy structure is not strict, we can’t use methods such as TopDown or MiddleOut. In this example we use BottomUp and MinTrace. If you want to calculate prediction intervals, you have to use the level argument as follows and set intervals_method='bootstrap'.

reconcilers = [
    BottomUp(),
    MinTrace(method='mint_shrink'),
    MinTrace(method='ols')
]
hrec = HierarchicalReconciliation(reconcilers=reconcilers)
Y_rec_df = hrec.reconcile(Y_hat_df=Y_hat_df, Y_df=Y_fitted_df, S=S_df, 
                          tags=tags, level=[80, 90], 
                          intervals_method='bootstrap')

The dataframe Y_rec_df contains the reconciled forecasts.

Y_rec_df.head()
dsETSETS/BottomUpETS/BottomUp-lo-90ETS/BottomUp-lo-80ETS/BottomUp-hi-80ETS/BottomUp-hi-90ETS/MinTrace_method-mint_shrinkETS/MinTrace_method-mint_shrink-lo-90ETS/MinTrace_method-mint_shrink-lo-80ETS/MinTrace_method-mint_shrink-hi-80ETS/MinTrace_method-mint_shrink-hi-90ETS/MinTrace_method-olsETS/MinTrace_method-ols-lo-90ETS/MinTrace_method-ols-lo-80ETS/MinTrace_method-ols-hi-80ETS/MinTrace_method-ols-hi-90
unique_id
Australia2016-01-0126080.87890624487.34960923244.12099623333.69472725381.79296925426.33398425532.52355924428.91170124709.21063826365.60693426476.25550126034.11424124914.13637525100.46293827102.73502227176.416922
Australia2016-04-0124587.01171923069.74414121826.51943421912.96289123946.60625024281.44726624118.55717723199.96862623295.24425225108.47041025489.38360624567.46099523484.05056823640.63842325709.76367825809.249492
Australia2016-07-0124147.30859422689.77734421297.13671921530.43828123701.17382824155.82031223731.25138722627.63966922818.72918224821.48845825246.86743224150.13489823030.15683423155.02555625359.99237625404.841402
Australia2016-10-0124794.04101623429.75781222037.12304722276.45312524241.41796924441.16015624486.54934423385.92723223600.70452525353.55562525481.47855724831.58451623725.92446423836.47517425900.20525425977.265089
Australia2017-01-0126284.00000024940.04296923696.72275423904.38281225814.94140625974.16992226041.86748824972.07785825158.98671026918.10474727135.58084526348.20333525254.65932425487.50229127410.87303527477.334507

​
Plot Predictions

Then we can plot the probabilist forecasts using the following function.

plot_df = pd.concat([Y_df.set_index(['unique_id', 'ds']), 
                     Y_rec_df.set_index('ds', append=True)], axis=1)
plot_df = plot_df.reset_index('ds')

​
Plot single time series

hplot.plot_series(
    series='Australia',
    Y_df=plot_df, 
    models=['y', 'ETS', 'ETS/MinTrace_method-ols', 'ETS/MinTrace_method-mint_shrink'],
    level=[80]
)


# Since we are plotting a bottom time series
# the probabilistic and mean forecasts
# differ due to bootstrapping
hplot.plot_series(
    series='Australia/Western Australia/Experience Perth/Visiting',
    Y_df=plot_df, 
    models=['y', 'ETS', 'ETS/BottomUp'],
    level=[80]
)

​
Plot hierarchichally linked time series

hplot.plot_hierarchically_linked_series(
    bottom_series='Australia/Western Australia/Experience Perth/Visiting',
    Y_df=plot_df, 
    models=['y', 'ETS', 'ETS/MinTrace_method-ols', 'ETS/BottomUp'],
    level=[80]
)


# ACT only has Canberra
hplot.plot_hierarchically_linked_series(
    bottom_series='Australia/ACT/Canberra/Other',
    Y_df=plot_df, 
    models=['y', 'ETS/MinTrace_method-mint_shrink'],
    level=[80, 90]
)

​
References

NormalityPERMBU