gluonts

Why we included this project

GluonTS is a Python library for building probabilistic forecasting models, with deep learning estimators implemented in PyTorch. It offers a set of neural forecasting approaches, including DeepAR and Transformer, behind a consistent estimator interface, so you can move from a CSV of historical values to forecast distributions in a few lines of code. The probabilistic output is the point: instead of a single point estimate, you get forecast distributions, which matters when business decisions need confidence bounds. Data scientists and ML engineers working on demand, capacity, or inventory planning will appreciate the clean PyTorch integration and the built-in evaluation utilities. It also works well as a baseline to compare against custom models or newer zero-shot approaches.