// the find
AutoViML/AutoViz
Automatically Visualize any dataset, any size with a single line of code. Created by Ram Seshadri. Collaborators Welcome. Permission Granted upon Request.
AutoViz generates exploratory data analysis charts from a dataframe or CSV with a single function call. It auto-detects column types, picks appropriate chart types, and supports matplotlib, Bokeh, and interactive server modes. Aimed at data scientists who want a quick first look at a new dataset without writing boilerplate plotting code.
Genuinely useful default behavior — it picks chart types based on variable types (continuous vs categorical vs target) rather than just dumping histograms. Multiple output backends (static PNG/SVG, interactive Bokeh, live server dashboard) cover most notebook and reporting workflows. The FixDQ() data quality companion is a practical addition that catches obvious problems before you start modeling. Sampling strategy for large datasets is sensible — statistically valid sample rather than just taking the first N rows.
The API design is showing its age: a single method with 10+ positional arguments and magic string dispatch for chart_format is hard to discover and easy to misuse. Three separate requirements files for different Python versions is a red flag — it means dependency management is manual and brittle rather than solved properly. Last meaningful activity was mid-2024 and the library has accumulated real drift against current pandas/matplotlib/Bokeh versions that users hit in the issue tracker. No programmatic access to the generated figures — you get side effects (charts displayed or saved) but can't easily grab a matplotlib Axes or Bokeh Figure to customize further.