AI4TS

0.0.3 verified Mon Apr 27 auth: no python

AI4TS (AI for Time Series) is a Python library for time series forecasting, classification, and anomaly detection using deep learning. Current version: 0.0.3, released 2023-11. The library is in early development with frequent releases.

pip install ai4ts

Common errors

error ModuleNotFoundError: No module named 'ai4ts' ↓

cause Library not installed or installed in wrong environment.

fix

Run pip install ai4ts in your active Python environment.

error ImportError: cannot import name 'TSDataset' from 'ai4ts' ↓

cause Incorrect import path; TSDataset is under ai4ts.datasets.

fix

Use from ai4ts.datasets import TSDataset.

Warnings

gotcha The library is in early alpha (v0.0.3). APIs are unstable and may change without notice. ↓

fix Pin to exact version and monitor GitHub for changes.

deprecated Some dataset loading functions use a different signature in GitHub examples vs. installed version. Check version compatibility. ↓

fix Refer to the documentation for your installed version; use `pip show ai4ts` to verify.

Imports

TSDataset

wrong

from ai4ts import TSDataset

correct

from ai4ts.datasets import TSDataset

TSDataset is under datasets submodule

TimeSeriesForecaster

wrong

from ai4ts import TimeSeriesForecaster

correct

from ai4ts.models import TimeSeriesForecaster

Model classes are under models submodule

Quickstart

Load a sample dataset and train a simple forecaster.

from ai4ts.datasets import TSDataset
from ai4ts.models import TimeSeriesForecaster
import torch

dataset = TSDataset()
X, y = dataset.load_data('airline')
model = TimeSeriesForecaster(input_size=1, hidden_size=64, num_layers=2)
model.train(X, y, epochs=10)