OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Tactis

TACTiS-2 is an open-source Python deep-learning model for probabilistic multivariate time-series forecasting and interpolation, including unaligned and unevenly sampled data.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Tactis — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Consider TACTiS-2 when building a Python workflow for multivariate forecasting or interpolation, particularly when the time series are not aligned or sampled at even intervals.

Who it is for

  • Researchers developing forecasting or interpolation workflows across multiple related time series.
  • Practitioners working with series whose observations are unaligned, unevenly sampled or incomplete.

Documented uses

  • Generate probabilistic forecasts for multiple related time series.
  • Interpolate observations across multivariate time series.
  • Train and evaluate a forecasting model on a selected dataset using the provided research materials.
Before you choose

Requirements & limits

The public repository lists an Apache-2.0 license. The model is installable as a Python package with pip; the repository notes testing with Python 3.10.8.

  • The standard package and the research extra are separate install choices; the research extra installs GluonTS and PyTorchTS for replicating paper experiments.
  • The training script exposes dataset, task and model hyperparameter options, so users need to supply a dataset and configure a training workflow.
  • The repository links a separate implementation of the original TACTiS model; the current README describes TACTiS-2.

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

Install TACTiS-2 for research workflows

Install the package variant that includes dependencies identified for reproducing the paper's experiments.

Before you start

  • Python environment; the README says tactis has been tested with Python 3.10.8.

Steps

  1. Install the research variant with the command below.

    Documented command
    pip install tactis[research]

  2. Use the included research code's train.py to specify a dataset, forecasting or interpolation task, hyperparameters and other training options.

Expected outcome

The research package and its GluonTS and PyTorchTS dependencies are installed, and the documented training script can be configured for a dataset and task.

Guide limitations

  • The README identifies GluonTS and PyTorchTS as dependencies of the research version; it does not provide dataset-specific instructions in the supplied material.
Documentation 1 ↗
Evidence and transparency

Sources & review

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