OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Data & Analytics

Orbit — uber

Orbit is an open-source Python library for Bayesian time-series forecasting and inference. It offers object-oriented model interfaces and an initialize–fit–predict workflow, with implementations for ETS, LGT, DLT and KTR and estimation options including MCMC, MAP and variational inference.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Orbit — uber — OpenViST project profile in AI Data & Analytics.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Orbit when you need model-based time-series forecasts and inference within Python, and want to work with its listed model families and estimation methods.

Who it is for

  • Python analysts forecasting dated data such as demand, sales or operational metrics.
  • Developers comparing probabilistic time-series models or incorporating forecasts into Python analysis workflows.

Documented uses

  • Fit a supported Bayesian time-series model to historical observations.
  • Generate forecasts and compare them with held-out actual data.
  • Compare model and estimation approaches for a forecasting task.
Before you choose

Requirements & limits

The project is open-source and its repository documents installation through PyPI with pip, conda-forge with conda, or from source.

  • Orbit is a Python library rather than a hosted forecasting application.
  • The documented model choices are ETS, LGT, DLT and KTR; choose among these based on the time-series task.
  • The README identifies cmdstanpy as a core dependency for Bayesian sampling.
  • The repository warns that new experimental code may have APIs subject to change and points stable-release users to the master branch.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Fit a DLT model and create predictions with the included claims dataset

Use Orbit's Damped Local Trend model to fit claims data and produce predictions for a held-out period.

Before you start

  • Orbit installed.
  • The example uses Orbit's load_iclaims dataset utility and the DLT model.

Steps

  1. Load the claims dataset with load_iclaims() and split off the final 52 rows as test data; use the earlier rows for training.

  2. Create a DLT model with response column claims, date column week, regressors trend.unemploy, trend.filling, and trend.job, and seasonality 52.

  3. Fit the model on the training dataframe with dlt.fit(df=train_df).

  4. Generate predictions for the test dataframe with dlt.predict(df=test_df) and plot them with plot_predicted_data, supplying the training, predicted, and test dataframes and the model's date and response columns.

Expected outcome

A predicted dataframe for the held-out period and a plot comparing predictions with training and test actuals.

Guide limitations

  • The demonstrated example is a DLT claims-data workflow with a 52-row holdout; it does not establish suitability for other datasets or model configurations.
  • The project notes that cmdstanpy is required as a core dependency for Bayesian sampling and that experimental APIs may change.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
uber/orbit · Official homepage title · retained in its source language
How we review sourcesSuggest a correction