Quick Start Tutorial#
GluonTS contains:
A number of pre-built models
Components for building new models (likelihoods, feature processing pipelines, calendar features etc.)
Data loading and processing
Plotting and evaluation facilities
Artificial and real datasets (only external datasets with blessed license)
[1]:
%matplotlib inline
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import json
Datasets#
Provided datasets#
GluonTS comes with a number of publicly available datasets.
[2]:
from gluonts.dataset.repository import get_dataset, dataset_names
from gluonts.dataset.util import to_pandas
[3]:
print(f"Available datasets: {dataset_names}")
Available datasets: ['constant', 'exchange_rate', 'solar-energy', 'electricity', 'traffic', 'exchange_rate_nips', 'electricity_nips', 'traffic_nips', 'solar_nips', 'wiki2000_nips', 'wiki-rolling_nips', 'taxi_30min', 'kaggle_web_traffic_with_missing', 'kaggle_web_traffic_without_missing', 'kaggle_web_traffic_weekly', 'm1_yearly', 'm1_quarterly', 'm1_monthly', 'nn5_daily_with_missing', 'nn5_daily_without_missing', 'nn5_weekly', 'tourism_monthly', 'tourism_quarterly', 'tourism_yearly', 'cif_2016', 'london_smart_meters_without_missing', 'wind_farms_without_missing', 'car_parts_without_missing', 'dominick', 'fred_md', 'pedestrian_counts', 'hospital', 'covid_deaths', 'kdd_cup_2018_without_missing', 'weather', 'm3_monthly', 'm3_quarterly', 'm3_yearly', 'm3_other', 'm4_hourly', 'm4_daily', 'm4_weekly', 'm4_monthly', 'm4_quarterly', 'm4_yearly', 'm5', 'uber_tlc_daily', 'uber_tlc_hourly', 'airpassengers', 'australian_electricity_demand', 'electricity_hourly', 'electricity_weekly', 'rideshare_without_missing', 'saugeenday', 'solar_10_minutes', 'solar_weekly', 'sunspot_without_missing', 'temperature_rain_without_missing', 'vehicle_trips_without_missing', 'ercot', 'ett_small_15min', 'ett_small_1h']
To download one of the built-in datasets, simply call get_dataset with one of the above names. GluonTS can re-use the saved dataset so that it does not need to be downloaded again the next time around.
[4]:
dataset = get_dataset("m4_hourly")
In general, the datasets provided by GluonTS are objects that consists of three main members:
dataset.trainis an iterable collection of data entries used for training. Each entry corresponds to one time series.dataset.testis an iterable collection of data entries used for inference. The test dataset is an extended version of the train dataset that contains a window in the end of each time series that was not seen during training. This window has length equal to the recommended prediction length.dataset.metadatacontains metadata of the dataset such as the frequency of the time series, a recommended prediction horizon, associated features, etc.
[5]:
entry = next(iter(dataset.train))
train_series = to_pandas(entry)
train_series.plot()
plt.grid(which="both")
plt.legend(["train series"], loc="upper left")
plt.show()
/opt/hostedtoolcache/Python/3.11.13/x64/lib/python3.11/site-packages/gluonts/dataset/common.py:263: FutureWarning: 'H' is deprecated and will be removed in a future version, please use 'h' instead.
return pd.Period(val, freq)
[6]:
entry = next(iter(dataset.test))
test_series = to_pandas(entry)
test_series.plot()
plt.axvline(train_series.index[-1], color="r") # end of train dataset
plt.grid(which="both")
plt.legend(["test series", "end of train series"], loc="upper left")
plt.show()
[7]:
print(
f"Length of forecasting window in test dataset: {len(test_series) - len(train_series)}"
)
print(f"Recommended prediction horizon: {dataset.metadata.prediction_length}")
print(f"Frequency of the time series: {dataset.metadata.freq}")
Length of forecasting window in test dataset: 48
Recommended prediction horizon: 48
Frequency of the time series: H
Custom datasets#
At this point, it is important to emphasize that GluonTS does not require this specific format for a custom dataset that a user may have. The only requirements for a custom dataset are to be iterable and have a “target” and a “start” field. To make this more clear, assume the common case where a dataset is in the form of a numpy.array and the index of the time series in a pandas.Period (possibly different for each time series):
[8]:
N = 10 # number of time series
T = 100 # number of timesteps
prediction_length = 24
freq = "1H"
custom_dataset = np.random.normal(size=(N, T))
start = pd.Period("01-01-2019", freq=freq) # can be different for each time series
/tmp/ipykernel_3977/2007592095.py:6: FutureWarning: 'H' is deprecated and will be removed in a future version, please use 'h' instead.
start = pd.Period("01-01-2019", freq=freq) # can be different for each time series
Now, you can split your dataset and bring it in a GluonTS appropriate format with just two lines of code:
[9]:
from gluonts.dataset.common import ListDataset
[10]:
# train dataset: cut the last window of length "prediction_length", add "target" and "start" fields
train_ds = ListDataset(
[{"target": x, "start": start} for x in custom_dataset[:, :-prediction_length]],
freq=freq,
)
# test dataset: use the whole dataset, add "target" and "start" fields
test_ds = ListDataset(
[{"target": x, "start": start} for x in custom_dataset], freq=freq
)
/opt/hostedtoolcache/Python/3.11.13/x64/lib/python3.11/site-packages/gluonts/dataset/common.py:255: FutureWarning: 'H' is deprecated and will be removed in a future version, please use 'h' instead.
ProcessDataEntry(to_offset(freq), one_dim_target, use_timestamp),
Training an existing model (Estimator)#
GluonTS comes with a number of pre-built models. All the user needs to do is configure some hyperparameters. The existing models focus on (but are not limited to) probabilistic forecasting. Probabilistic forecasts are predictions in the form of a probability distribution, rather than simply a single point estimate.
We will begin with GluonTS’s pre-built feedforward neural network estimator, a simple but powerful forecasting model. We will use this model to demonstrate the process of training a model, producing forecasts, and evaluating the results.
GluonTS’s built-in feedforward neural network (SimpleFeedForwardEstimator) accepts an input window of length context_length and predicts the distribution of the values of the subsequent prediction_length values. In GluonTS parlance, the feedforward neural network model is an example of an Estimator. In GluonTS, Estimator objects represent a forecasting model as well as details such as its coefficients, weights, etc.
In general, each estimator (pre-built or custom) is configured by a number of hyperparameters that can be either common (but not binding) among all estimators (e.g., the prediction_length) or specific for the particular estimator (e.g., number of layers for a neural network or the stride in a CNN).
Finally, each estimator is configured by a Trainer, which defines how the model will be trained i.e., the number of epochs, the learning rate, etc.
[11]:
from gluonts.mx import SimpleFeedForwardEstimator, Trainer
[12]:
estimator = SimpleFeedForwardEstimator(
num_hidden_dimensions=[10],
prediction_length=dataset.metadata.prediction_length,
context_length=100,
trainer=Trainer(ctx="cpu", epochs=5, learning_rate=1e-3, num_batches_per_epoch=100),
)
After specifying our estimator with all the necessary hyperparameters we can train it using our training dataset dataset.train by invoking the train method of the estimator. The training algorithm returns a fitted model (or a Predictor in GluonTS parlance) that can be used to construct forecasts.
[13]:
predictor = estimator.train(dataset.train)
100%|██████████| 100/100 [00:00<00:00, 151.80it/s, epoch=1/5, avg_epoch_loss=5.64]
100%|██████████| 100/100 [00:00<00:00, 150.09it/s, epoch=2/5, avg_epoch_loss=4.89]
100%|██████████| 100/100 [00:00<00:00, 153.50it/s, epoch=3/5, avg_epoch_loss=4.73]
100%|██████████| 100/100 [00:00<00:00, 152.73it/s, epoch=4/5, avg_epoch_loss=4.71]
100%|██████████| 100/100 [00:00<00:00, 150.61it/s, epoch=5/5, avg_epoch_loss=4.83]
Visualize and evaluate forecasts#
With a predictor in hand, we can now predict the last window of the dataset.test and evaluate our model’s performance.
GluonTS comes with the make_evaluation_predictions function that automates the process of prediction and model evaluation. Roughly, this function performs the following steps:
Removes the final window of length
prediction_lengthof thedataset.testthat we want to predictThe estimator uses the remaining data to predict (in the form of sample paths) the “future” window that was just removed
The module outputs the forecast sample paths and the
dataset.test(as python generator objects)
[14]:
from gluonts.evaluation import make_evaluation_predictions
[15]:
forecast_it, ts_it = make_evaluation_predictions(
dataset=dataset.test, # test dataset
predictor=predictor, # predictor
num_samples=100, # number of sample paths we want for evaluation
)
First, we can convert these generators to lists to ease the subsequent computations.
[16]:
forecasts = list(forecast_it)
tss = list(ts_it)
We can examine the first element of these lists (that corresponds to the first time series of the dataset). Let’s start with the list containing the time series, i.e., tss. We expect the first entry of tss to contain the (target of the) first time series of dataset.test.
[17]:
# first entry of the time series list
ts_entry = tss[0]
[18]:
# first 5 values of the time series (convert from pandas to numpy)
np.array(ts_entry[:5]).reshape(
-1,
)
[18]:
array([605., 586., 586., 559., 511.], dtype=float32)
[19]:
# first entry of dataset.test
dataset_test_entry = next(iter(dataset.test))
[20]:
# first 5 values
dataset_test_entry["target"][:5]
[20]:
array([605., 586., 586., 559., 511.], dtype=float32)
The entries in the forecast list are a bit more complex. They are objects that contain all the sample paths in the form of numpy.ndarray with dimension (num_samples, prediction_length), the start date of the forecast, the frequency of the time series, etc. We can access all this information by simply invoking the corresponding attribute of the forecast object.
[21]:
# first entry of the forecast list
forecast_entry = forecasts[0]
[22]:
print(f"Number of sample paths: {forecast_entry.num_samples}")
print(f"Dimension of samples: {forecast_entry.samples.shape}")
print(f"Start date of the forecast window: {forecast_entry.start_date}")
print(f"Frequency of the time series: {forecast_entry.freq}")
Number of sample paths: 100
Dimension of samples: (100, 48)
Start date of the forecast window: 1750-01-30 04:00
Frequency of the time series: <Hour>
We can also do calculations to summarize the sample paths, such as computing the mean or a quantile for each of the 48 time steps in the forecast window.
[23]:
print(f"Mean of the future window:\n {forecast_entry.mean}")
print(f"0.5-quantile (median) of the future window:\n {forecast_entry.quantile(0.5)}")
Mean of the future window:
[656.8301 593.62317 507.7621 560.19147 533.6132 477.68454 475.19266
521.62006 559.07806 607.34045 507.19827 617.89185 700.01794 812.90155
863.4309 844.4272 766.49335 811.2046 819.93756 871.2954 777.4522
842.5376 778.62775 759.9309 659.16516 535.54474 585.18896 425.0777
511.04825 451.5377 471.60297 510.15735 512.3272 552.13367 589.8035
733.70197 802.0394 818.2543 832.368 877.9747 965.2444 829.30585
927.90326 875.8522 793.9871 756.0211 797.9769 690.6666 ]
0.5-quantile (median) of the future window:
[672.58746 586.11975 519.2786 564.1543 515.0018 484.15073 492.16486
528.90814 561.7334 601.94604 500.9475 613.4544 713.8728 819.8752
862.93646 852.6691 800.42615 822.4439 823.6065 877.26666 798.9276
822.2547 780.85736 749.40216 658.12726 547.11584 583.4626 439.69043
525.48395 434.80704 470.32306 501.18228 503.32404 553.00684 581.06006
724.60266 784.80896 820.8987 822.7163 900.4033 966.26355 843.14075
925.8449 851.0002 776.1264 783.4757 815.173 681.5972 ]
Forecast objects have a plot method that can summarize the forecast paths as the mean, prediction intervals, etc. The prediction intervals are shaded in different colors as a “fan chart”.
[24]:
plt.plot(ts_entry[-150:].to_timestamp())
forecast_entry.plot(show_label=True)
plt.legend()
[24]:
<matplotlib.legend.Legend at 0x7fec65672c50>
We can also evaluate the quality of our forecasts numerically. In GluonTS, the Evaluator class can compute aggregate performance metrics, as well as metrics per time series (which can be useful for analyzing performance across heterogeneous time series).
[25]:
from gluonts.evaluation import Evaluator
[26]:
evaluator = Evaluator(quantiles=[0.1, 0.5, 0.9])
agg_metrics, item_metrics = evaluator(tss, forecasts)
Running evaluation: 414it [00:00, 11254.99it/s]
The aggregate metrics, agg_metrics, aggregate both across time-steps and across time series.
[27]:
print(json.dumps(agg_metrics, indent=4))
{
"MSE": 10353652.54473964,
"abs_error": 10360954.195072174,
"abs_target_sum": 145558863.59960938,
"abs_target_mean": 7324.822041043146,
"seasonal_error": 336.9046924038305,
"MASE": 3.9537099803003812,
"MAPE": 0.2557887818346373,
"sMAPE": 0.19288036810315173,
"MSIS": 61.65597012019594,
"num_masked_target_values": 0.0,
"QuantileLoss[0.1]": 4983550.784189797,
"Coverage[0.1]": 0.10451892109500804,
"QuantileLoss[0.5]": 10360954.240943432,
"Coverage[0.5]": 0.5307467793880837,
"QuantileLoss[0.9]": 6920760.359059906,
"Coverage[0.9]": 0.8778683574879227,
"RMSE": 3217.7092076102276,
"NRMSE": 0.43928837991973735,
"ND": 0.07118051033685027,
"wQuantileLoss[0.1]": 0.034237357045450105,
"wQuantileLoss[0.5]": 0.07118051065198916,
"wQuantileLoss[0.9]": 0.04754612799188189,
"mean_absolute_QuantileLoss": 7421755.128064379,
"mean_wQuantileLoss": 0.05098799856310706,
"MAE_Coverage": 0.3926227858293075,
"OWA": NaN
}
Individual metrics are aggregated only across time-steps.
[28]:
item_metrics.head()
[28]:
| item_id | forecast_start | MSE | abs_error | abs_target_sum | abs_target_mean | seasonal_error | MASE | MAPE | sMAPE | num_masked_target_values | ND | MSIS | QuantileLoss[0.1] | Coverage[0.1] | QuantileLoss[0.5] | Coverage[0.5] | QuantileLoss[0.9] | Coverage[0.9] | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 1750-01-30 04:00 | 3388.305664 | 2193.886230 | 31644.0 | 659.250000 | 42.371302 | 1.078701 | 0.069746 | 0.067837 | 0.0 | 0.069330 | 13.774889 | 1029.080524 | 0.020833 | 2193.886322 | 0.770833 | 1439.796960 | 1.000000 |
| 1 | 1 | 1750-01-30 04:00 | 174607.135417 | 18427.251953 | 124149.0 | 2586.437500 | 165.107988 | 2.325151 | 0.156027 | 0.142711 | 0.0 | 0.148429 | 13.605702 | 4572.551636 | 0.312500 | 18427.251831 | 0.979167 | 8531.256982 | 1.000000 |
| 2 | 2 | 1750-01-30 04:00 | 36602.427083 | 7121.232422 | 65030.0 | 1354.791667 | 78.889053 | 1.880603 | 0.100835 | 0.107245 | 0.0 | 0.109507 | 13.800707 | 3489.670313 | 0.000000 | 7121.232178 | 0.187500 | 2419.273828 | 0.770833 |
| 3 | 3 | 1750-01-30 04:00 | 223569.291667 | 18354.597656 | 235783.0 | 4912.145833 | 258.982249 | 1.476501 | 0.079401 | 0.079289 | 0.0 | 0.077845 | 14.135824 | 9796.419873 | 0.062500 | 18354.597656 | 0.375000 | 7536.722900 | 0.979167 |
| 4 | 4 | 1750-01-30 04:00 | 103810.302083 | 11672.973633 | 131088.0 | 2731.000000 | 200.494083 | 1.212938 | 0.089112 | 0.084565 | 0.0 | 0.089047 | 13.218125 | 5095.128149 | 0.062500 | 11672.973755 | 0.666667 | 7300.917358 | 1.000000 |
[29]:
item_metrics.plot(x="MSIS", y="MASE", kind="scatter")
plt.grid(which="both")
plt.show()
