AutoGluon Time Series - Forecasting Quick Start

Via a simple fit() call, AutoGluon can train and tune

• simple forecasting models (e.g., ARIMA, ETS, Theta),

• powerful deep learning models (e.g., DeepAR, Temporal Fusion Transformer),

• tree-based models (e.g., LightGBM),

• an ensemble that combines predictions of other models

to produce multi-step ahead probabilistic forecasts for univariate time series data.

This tutorial demonstrates how to quickly start using AutoGluon to generate hourly forecasts for the M4 forecasting competition dataset.

Loading time series data as a TimeSeriesDataFrame

First, we import some required modules

import pandas as pd
from autogluon.timeseries import TimeSeriesDataFrame, TimeSeriesPredictor

To use autogluon.timeseries, we will only need the following two classes:

• TimeSeriesDataFrame stores a dataset consisting of multiple time series.

• TimeSeriesPredictor takes care of fitting, tuning and selecting the best forecasting models, as well as generating new forecasts.

We load a subset of the M4 hourly dataset as a pandas.DataFrame

df = pd.read_csv("https://autogluon.s3.amazonaws.com/datasets/timeseries/m4_hourly_subset/train.csv")
df.head()

	item_id	timestamp	target
0	H1	1750-01-01 00:00:00	605.0
1	H1	1750-01-01 01:00:00	586.0
2	H1	1750-01-01 02:00:00	586.0
3	H1	1750-01-01 03:00:00	559.0
4	H1	1750-01-01 04:00:00	511.0

AutoGluon expects time series data in long format. Each row of the data frame contains a single observation (timestep) of a single time series represented by

• unique ID of the time series ("item_id") as int or str

• timestamp of the observation ("timestamp") as a pandas.Timestamp or compatible format

• numeric value of the time series ("target")

The raw dataset should always follow this format with at least three columns for unique ID, timestamp, and target value, but the names of these columns can be arbitrary. It is important, however, that we provide the names of the columns when constructing a TimeSeriesDataFrame that is used by AutoGluon. AutoGluon will raise an exception if the data doesn’t match the expected format.

train_data = TimeSeriesDataFrame.from_data_frame(
    df,
    id_column="item_id",
    timestamp_column="timestamp"
)
train_data.head()

		target
item_id	timestamp
H1	1750-01-01 00:00:00	605.0
	1750-01-01 01:00:00	586.0
	1750-01-01 02:00:00	586.0
	1750-01-01 03:00:00	559.0
	1750-01-01 04:00:00	511.0

We refer to each individual time series stored in a TimeSeriesDataFrame as an item. For example, items might correspond to different products in demand forecasting, or to different stocks in financial datasets. This setting is also referred to as a panel of time series. Note that this is not the same as multivariate forecasting — AutoGluon generates forecasts for each time series individually, without modeling interactions between different items (time series).

TimeSeriesDataFrame inherits from pandas.DataFrame, so all attributes and methods of pandas.DataFrame are available in a TimeSeriesDataFrame. It also provides other utility functions, such as loaders for different data formats (see TimeSeriesDataFrame for details).

Training time series models with TimeSeriesPredictor.fit

To forecast future values of the time series, we need to create a TimeSeriesPredictor object.

Models in autogluon.timeseries forecast time series multiple steps into the future. We choose the number of these steps — the prediction length (also known as the forecast horizon) — depending on our task. For example, our dataset contains time series measured at hourly frequency, so we set prediction_length = 48 to train models that forecast up to 48 hours into the future.

We instruct AutoGluon to save trained models in the folder ./autogluon-m4-hourly. We also specify that AutoGluon should rank models according to mean absolute scaled error (MASE), and that data that we want to forecast is stored in the column "target" of the TimeSeriesDataFrame.

predictor = TimeSeriesPredictor(
    prediction_length=48,
    path="autogluon-m4-hourly",
    target="target",
    eval_metric="MASE",
)

predictor.fit(
    train_data,
    presets="medium_quality",
    time_limit=600,
)

================ TimeSeriesPredictor ================
TimeSeriesPredictor.fit() called
Setting presets to: medium_quality
Fitting with arguments:
{'enable_ensemble': True,
 'evaluation_metric': 'MASE',
 'excluded_model_types': None,
 'hyperparameter_tune_kwargs': None,
 'hyperparameters': 'medium_quality',
 'num_val_windows': 1,
 'prediction_length': 48,
 'random_seed': None,
 'target': 'target',
 'time_limit': 600,
 'verbosity': 2}
Provided training data set with 148060 rows, 200 items (item = single time series). Average time series length is 740.3. Data frequency is 'H'.
=====================================================
AutoGluon will save models to autogluon-m4-hourly/
AutoGluon will gauge predictive performance using evaluation metric: 'MASE'
	This metric's sign has been flipped to adhere to being 'higher is better'. The reported score can be multiplied by -1 to get the metric value.

Provided dataset contains following columns:
	target:           'target'

Starting training. Start time is 2023-06-29 22:42:13
Models that will be trained: ['Naive', 'SeasonalNaive', 'Theta', 'AutoETS', 'RecursiveTabular', 'DeepAR']
Training timeseries model Naive. Training for up to 599.65s of the 599.65s of remaining time.
	-6.6629       = Validation score (-MASE)
	0.12    s     = Training runtime
	4.23    s     = Validation (prediction) runtime
Training timeseries model SeasonalNaive. Training for up to 595.30s of the 595.30s of remaining time.
	-1.2169       = Validation score (-MASE)
	0.11    s     = Training runtime
	0.22    s     = Validation (prediction) runtime
Training timeseries model Theta. Training for up to 594.96s of the 594.96s of remaining time.
	-2.1425       = Validation score (-MASE)
	0.11    s     = Training runtime
	28.13   s     = Validation (prediction) runtime
Training timeseries model AutoETS. Training for up to 566.70s of the 566.70s of remaining time.
	-1.9399       = Validation score (-MASE)
	0.11    s     = Training runtime
	101.97  s     = Validation (prediction) runtime
Training timeseries model RecursiveTabular. Training for up to 464.62s of the 464.62s of remaining time.
	-0.8988       = Validation score (-MASE)
	14.27   s     = Training runtime
	2.44    s     = Validation (prediction) runtime
Training timeseries model DeepAR. Training for up to 447.88s of the 447.88s of remaining time.
	-1.8848       = Validation score (-MASE)
	94.43   s     = Training runtime
	2.04    s     = Validation (prediction) runtime
Fitting simple weighted ensemble.
	-0.8850       = Validation score (-MASE)
	6.11    s     = Training runtime
	106.67  s     = Validation (prediction) runtime
Training complete. Models trained: ['Naive', 'SeasonalNaive', 'Theta', 'AutoETS', 'RecursiveTabular', 'DeepAR', 'WeightedEnsemble']
Total runtime: 254.80 s
Best model: WeightedEnsemble
Best model score: -0.8850

<autogluon.timeseries.predictor.TimeSeriesPredictor at 0x7fbf2e8788e0>

Here we used the "medium_quality" presets and limited the training time to 10 minutes (600 seconds). The presets define which models AutoGluon will try to fit. For medium_quality presets, these are simple baselines (Naive, SeasonalNaive), statistical models (AutoETS, Theta), tree-based model LightGBM wrapped by RecursiveTabular, a deep learning model DeepAR, and a weighted ensemble combining these. Other available presets for TimeSeriesPredictor are "fast_training", "high_quality" and "best_quality". Higher quality presets will usually produce more accurate forecasts but take longer to train.

Inside fit(), AutoGluon will train as many models as possible within the given time limit. Trained models are then ranked based on their performance on an internal validation set. By default, this validation set is constructed by holding out the last prediction_length timesteps of each time series in train_data.

Generating forecasts with TimeSeriesPredictor.predict

We can now use the fitted TimeSeriesPredictor to forecast the future time series values. By default, AutoGluon will make forecasts using the model that had the best score on the internal validation set. The forecast always includes predictions for the next prediction_length timesteps, starting from the end of each time series in train_data.

predictions = predictor.predict(train_data)
predictions.head()

Global seed set to 123
Model not specified in predict, will default to the model with the best validation score: WeightedEnsemble

		mean	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
item_id	timestamp
H1	1750-01-30 04:00:00	623.826478	592.187364	603.071471	610.896354	617.586165	623.839536	630.088389	636.772222	644.605014	655.454230
	1750-01-30 05:00:00	557.547544	514.678373	529.443829	540.037456	549.108091	557.563545	566.026309	575.078872	585.677181	600.402671
	1750-01-30 06:00:00	514.927764	463.480392	481.166716	493.907031	504.765269	514.936364	525.100682	535.968523	548.697259	566.357110
	1750-01-30 07:00:00	480.985158	422.308391	442.465546	457.012861	469.396393	480.990611	492.562915	504.980224	519.509260	539.677283
	1750-01-30 08:00:00	458.342474	393.299979	415.597355	431.712749	445.475217	458.345099	471.220778	484.969452	501.030328	523.435727

AutoGluon produces a probabilistic forecast: in addition to predicting the mean (expected value) of the time series in the future, models also provide the quantiles of the forecast distribution. The quantile forecasts give us an idea about the range of possible outcomes. For example, if the "0.1" quantile is equal to 500.0, it means that the model predicts a 10% chance that the target value will be below 500.0.

We will now visualize the forecast and the actually observed values for one of the time series in the dataset. We plot the mean forecast, as well as the 10% and 90% quantiles to show the range of potential outcomes.

import matplotlib.pyplot as plt

# TimeSeriesDataFrame can also be loaded directly from a file
test_data = TimeSeriesDataFrame.from_path("https://autogluon.s3.amazonaws.com/datasets/timeseries/m4_hourly_subset/test.csv")

plt.figure(figsize=(20, 3))

item_id = "H1"
y_past = train_data.loc[item_id]["target"]
y_pred = predictions.loc[item_id]
y_test = test_data.loc[item_id]["target"][-48:]

plt.plot(y_past[-200:], label="Past time series values")
plt.plot(y_pred["mean"], label="Mean forecast")
plt.plot(y_test, label="Future time series values")

plt.fill_between(
    y_pred.index, y_pred["0.1"], y_pred["0.9"], color="red", alpha=0.1, label=f"10%-90% confidence interval"
)
plt.legend();

Evaluating the performance of different models

We can view the performance of each model AutoGluon has trained via the leaderboard() method. We provide the test data set to the leaderboard function to see how well our fitted models are doing on the unseen test data. The leaderboard also includes the validation scores computed on the internal validation dataset.

In AutoGluon leaderboards, higher scores always correspond to better predictive performance. Therefore our MASE scores are multiplied by -1, such that higher “negative MASE”s correspond to more accurate forecasts.

# The test score is computed using the last
# prediction_length=48 timesteps of each time series in test_data
predictor.leaderboard(test_data, silent=True)

Additional data provided, testing on additional data. Resulting leaderboard will be sorted according to test score (`score_test`).

	model	score_test	score_val	pred_time_test	pred_time_val	fit_time_marginal	fit_order
0	WeightedEnsemble	-0.850541	-0.885005	100.841253	106.672521	6.112758	7
1	RecursiveTabular	-0.870271	-0.898770	1.789228	2.443860	14.273363	5
2	SeasonalNaive	-1.022854	-1.216909	0.237972	0.218211	0.114915	2
3	AutoETS	-1.778531	-1.939939	96.717588	101.967237	0.113186	4
4	DeepAR	-1.819948	-1.884837	2.049753	2.043213	94.428325	6
5	Theta	-1.905365	-2.142531	2.995191	28.133277	0.114513	3
6	Naive	-6.696079	-6.662942	0.224626	4.226221	0.118196	1

Summary

We used autogluon.timeseries to make probabilistic multi-step forecasts on the M4 Hourly dataset. Check out Forecasting Time Series - In Depth to learn about the advanced capabilities of AutoGluon for time series forecasting.