Time Series Forecasting Experiments

In this notebook we will show how to run a simple time series forecasting experiment using the tsml-eval package. Time series forecasting is the task of predicting future values of a time series.

time series classification ​

[ ]:

import numpy as np
from aeon.datasets import load_airline
from aeon.forecasting.model_selection import temporal_train_test_split

from tsml_eval.evaluation.storage import load_forecaster_results
from tsml_eval.experiments import (
    experiments,
    get_forecaster_by_name,
    run_forecasting_experiment,
)

[2]:

y = load_airline()
train, test = temporal_train_test_split(y)

[3]:

# set_regressor can be used to find various regressors by string, but
# any aeon, tsml or sklearn regressor can be used in the experiments function
forecaster = get_forecaster_by_name("NaiveForecaster")

# record memory usage every 0.1 seconds, just here for notebook speed
# does not need to be changed for usage
experiments.MEMRECORD_INTERVAL = 0.1

run_forecasting_experiment(
    np.array(train),
    np.array(test),
    forecaster,
    "./generated_results/",
    dataset_name="Airline",
    random_seed=0,
)

A function is also available to load the dataset as well as run an experiment, see load_and_run_forecasting_experiment in tsml_eval.experiments.

Both experiment functions will output a results file in the {results_dir}/{forecaster_name}/Predictions/{dataset_name}/ directory. These files can be loaded individually, or used as a collection in the evaluation module. See the evaluation notebook for more details.

[4]:

fr = load_forecaster_results(
    "./generated_results/NaiveForecaster/Predictions/Airline/testResample0.csv"
)
print(fr.predictions)
print(fr.mean_absolute_percentage_error)

[336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336.
 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336. 336.
 336. 336. 336. 336. 336. 336. 336. 336.]
0.19886711926999853

Generated using nbsphinx. The Jupyter notebook can be found here. binder