Nixtla    

Neural 🧠 Forecast

Deep Learning for time series

State-of-the-art time series forecasting for PyTorch.

NeuralForecast is a Python library for time series forecasting with deep learning models. It includes benchmark datasets, data-loading utilities, evaluation functions, statistical tests, univariate model benchmarks and SOTA models implemented in PyTorch and PyTorchLightning.

Getting started • Installation • Models

🎉 New!

• MQN-HiTS example: produce accurate and efficient probabilistic forecasts in long-horizon settings. Outperforming AutoARIMA's accuracy in a fraction of the time.
• N-HiTS example: load, train, and tune hyperparameter, to achieve SoTA. Outperform Transformers by 25% in 50x less time.

⚡ Why Deep Learning on Time Series?

Accuracy:

• Global model is fitted simultaneously for several time series.
• Shared information helps with highly parametrized and flexible models.
• Useful for items/skus that have little to no history available.

Efficiency:

• Automatic featurization processes.
• Fast computations (GPU or TPU).

📖 Documentation

Here is a link to the documentation.

🧬 Getting Started

Example Jupyter Notebook

💻 Installation

PyPI

You can install the released version of NeuralForecast from the Python package index with:

pip install neuralforecast

(Installing inside a python virtualenvironment or a conda environment is recommended.)

Conda

Also you can install the released version of NeuralForecast from conda with:

conda install -c conda-forge neuralforecast

(Installing inside a python virtualenvironment or a conda environment is recommended.)

Dev Mode

If you want to make some modifications to the code and see the effects in real time (without reinstalling), follow the steps below:

git clone https://github.com/Nixtla/neuralforecast.git
cd neuralforecast
pip install -e .

Forecasting models

• Neural Hierarchical Interpolation for Time Series Forecasting (N-HiTS): A new model for long-horizon forecasting which incorporates novel hierarchical interpolation and multi-rate data sampling techniques to specialize blocks of its architecture to different frequency band of the time-series signal. It achieves SoTA performance on several benchmark datasets, outperforming current Transformer-based models by more than 25%.

• Exponential Smoothing Recurrent Neural Network (ES-RNN): A hybrid model that combines the expressivity of non linear models to capture the trends while it normalizes using a Holt-Winters inspired model for the levels and seasonals. This model is the winner of the M4 forecasting competition.

• Neural Basis Expansion Analysis (N-BEATS): A model from Element-AI (Yoshua Bengio’s lab) that has proven to achieve state-of-the-art performance on benchmark large scale forecasting datasets like Tourism, M3, and M4. The model is fast to train and has an interpretable configuration.

• Neural Basis Expansion Analysis with Exogenous Variables (N-BEATSx): The neural basis expansion with exogenous variables is an extension to the original N-BEATS that allows it to include time dependent covariates.

• Transformer-Based Models: Transformer-based framework for unsupervised representation learning of multivariate time series.
  • Autoformer: Encoder-decoder model with decomposition capabilities and an approximation to attention based on Fourier transform.
  • Informer: Transformer with MLP based multi-step prediction strategy, that approximates self-attention with sparsity.
  • Transformer: Classical vanilla Transformer.

📃 License

This project is licensed under the GPLv3 License - see the LICENSE file for details.

🔨 How to contribute

See CONTRIBUTING.md.

Contributors ✨

Thanks goes to these wonderful people (emoji key):

fede 💻 🐛 📖	Greg DeVos 🤔	Cristian Challu 💻	mergenthaler 📖 💻	Kin 💻 🐛 🔣	José Morales 💻	Alejandro 💻
stefanialvs 🎨	Ikko Ashimine 🐛	vglaucus 🐛	Pietro Monticone 🐛

This project follows the all-contributors specification. Contributions of any kind welcome!