autogluon_timeseries_forecasting_with_confidence_intervals_visualization.py

1import pandas as pd
2from autogluon.timeseries import TimeSeriesDataFrame, TimeSeriesPredictor
3
4# 1. Load data from a CSV file or a Pandas DataFrame
5df = pd.read_csv("https://autogluon.s3.amazonaws.com/datasets/timeseries/m4_hourly_subset/train.csv")
6train_data = TimeSeriesDataFrame.from_path("https://autogluon.s3.amazonaws.com/datasets/timeseries/m4_hourly_subset/train.csv")
7
8# 2. Initialize the Predictor
9predictor = TimeSeriesPredictor(
10    prediction_length=24,
11    target="target",
12    eval_metric="MASE",
13)
14
15# 3. Train the models
16predictor.fit(
17    train_data,
18    presets="medium_quality",
19    time_limit=600,
20)
21
22# 4. Generate predictions
23predictions = predictor.predict(train_data)
24
25# 5. Visualize predictions (optional)
26import matplotlib.pyplot as plt
27
28# Select a specific time series ID to plot
29item_id = "H1"
30plt.figure(figsize=(10, 3))
31plt.plot(train_data.loc[item_id], label="Observed")
32plt.plot(predictions.loc[item_id]["mean"], label="Predicted")
33plt.fill_between(
34    predictions.loc[item_id].index,
35    predictions.loc[item_id]["0.1"],
36    predictions.loc[item_id]["0.9"],
37    color="red",
38    alpha=0.2,
39    label="Confidence Interval"
40)
41plt.legend()
42plt.show()