DSIPTS Documentation

DSIPTS (Deep Learning for Time Series) is a Python library for time series forecasting that allows you to train state-of-the-art deep learning models on your time series data or on benchmark datasets from the literature.

Note

This project is under active development. Contributions and feedback are welcome!

Key Features

  • 🚀 State-of-the-art Models: Includes Transformer-based models (Informer, Autoformer, PatchTST, iTransformer), convolutional models, RNNs, and more

  • 📊 Flexible Data Handling: Support for multi-variate time series with D1/D2 layer architecture

  • PyTorch Lightning Integration: Built on PyTorch Lightning for scalable training

  • 🔧 Easy Customization: Add your own architectures and compare against existing models

  • 📈 Benchmark Datasets: Built-in support for popular time series benchmarks

Getting Started

User Guide

Reference

API Reference

Examples

Examples & Tutorials

Development

Development

Indices and Tables