Practical overview
When it fits
Use Lida when you want a language-model-assisted workflow for moving from a tabular dataset to visualization ideas and charts, or when you need to edit or explain the resulting visualization code. Consider preprocessing datasets with more than ten columns, and use the infographic path only with its beta status and extra dependencies in mind.
Who it is for
- Analysts exploring datasets and generating charts from candidate questions or goals.
- Developers building data-visualization workflows or apps with a Python library or optional web API.
- Notebook users who want to experiment with language-model-assisted visualization.
Documented uses
- Summarize a CSV or JSON dataset that can be loaded into a pandas dataframe.
- Generate visualization goals from a data summary, optionally using a persona.
- Generate charts with supported visualization libraries such as matplotlib, seaborn, Altair or D3.
- Edit, explain, evaluate, repair or recommend visualizations through code-oriented workflows.
- Generate data-faithful infographics using the separate beta feature.
Requirements & limits
The repository is public and identifies the project as MIT-licensed. The documented setup uses Python 3.10 or later and a configured language-model provider; the OpenAI example requires an API key.
- The documented input examples are CSV and JSON data loadable as a pandas dataframe.
- The project says it works best with datasets of ten or fewer columns; larger datasets may require selecting and preprocessing columns.
- Lida generates and executes visualization code, so its documentation recommends a secure environment with appropriate permissions.
- Infographic generation is marked beta and requires installing
lida[infographics]and its additional dependencies. - The project lists OpenAI, Azure OpenAI, PaLM, Cohere and Hugging Face among supported model providers; local Hugging Face use may require a GPU.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install Lida and prepare its model access
Set up the library and configure an OpenAI API key for its documented OpenAI-based workflow.
Before you start
- Python 3.10 or later
- pip
- An OpenAI API key
Steps
Install Lida with the command below.
pip install -U lidaUpdate its documented dependencies with the command below.
pip install -U llmx openaiSet the
OPENAI_API_KEYenvironment variable to your key.
Expected outcome
The documented Python environment and OpenAI key are configured for trying Lida through its tutorial notebook or bundled UI.
Guide limitations
- This covers setup only; it does not verify model access or chart generation.
- Lida generates and executes visualization code; its documentation advises using a secure environment with appropriate permissions.
Start Lida’s bundled web UI and API
Open the locally bundled interface and view its API documentation after installing and configuring Lida.
Before you start
- Lida installed in a Python 3.10-or-later environment
- An OpenAI API key set as
OPENAI_API_KEY
Steps
Run
lida ui --port=8080 --docs.Open
http://localhost:8080/for the UI.Open
http://localhost:8080/api/docsfor the API specification.
Expected outcome
The README’s documented commands provide local URLs for the bundled UI and API documentation.
Guide limitations
- This guide describes the documented startup path, not a verified deployment or execution test.
- Lida generates and executes visualization code, so use a secure environment with appropriate permissions.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- microsoft/lida · Official homepage title · retained in its source language