Lux: A Python API for Intelligent Visual Discovery
- Audience level:
Despite extensive charting support in the Jupyter ecosystem, few visualization libraries are designed to be conducive for ad-hoc data exploration. Lux is a lightweight API that automates aspects of data exploration, by recommending visualization in-situ user's Jupyter notebooks. The Lux Jupyter widget provides a seamless, flexible transition between code and interaction to advance exploration.
The objective of the poster is to introduce the Lux system and gather feedback from the Jupyter community. The talk is aimed at data science practitioners, although no prior background knowledge is required. The poster presentation will include video demos of key features.
The outline of the talk is as follows:
- Visualization allows users to explore and discover trends and patterns in their data. However, existing charting libraries in the Jupyter ecosystem are not conducive to exploration, as they require users to exactly specify what aspects of the data to visualize and how they want to visualize their data.
- Lux is a lightweight API that closes the gap between reasoning and execution by automating away the tedious choices required for generating exploratory visualizations.
- Key features:
- Lux recommends visualizations essentially for free to users by helping them visualize their Pandas dataframe, in-situ their Jupyter notebooks.
- Lux is integrated with an interactive Jupyter widget that allows users to quickly switch from programmatic specification to interactive browsing and vice versa.
- Lux offers a powerful, intuitive language for generating quick visualizations on-demand and for working with large collections of visualizations.
- By bringing the power of visualization recommendation directly into Jupyter notebooks, Lux facilitate faster experimentation with data and encourages users to visualize their data alongside other common data science tasks, such as data cleaning and modeling.