Details

The AutoViz architecture is designed as a streamlined data visualization pipeline, centered around the AutoViz API & Orchestrator which acts as a facade for automated exploratory data analysis. Data flows sequentially, beginning with the Data Loader & Classifier for ingestion and type inference, then moving to Data Preprocessing & Feature Engineering for refinement. Text-specific data is routed through the NLP Data Processor. Depending on the visualization requirements, processed data is fed into either the Matplotlib/Seaborn Plotting Engine for static outputs or the HoloViews/hvPlot Plotting Engine for interactive visualizations. All generated plots converge at the Visualization Output & Export Manager for final display and saving, providing a comprehensive and automated EDA workflow.

AutoViz API & Orchestrator

The primary user-facing interface and control center, orchestrating the entire EDA and visualization pipeline.

Data Loader & Classifier

Handles initial data loading and automatic classification of column types.

Data Preprocessing & Feature Engineering

Manages advanced data preparation, including cleaning, transformation, and feature selection.

NLP Data Processor

Specialized module for text data cleaning and text-specific visualization generation.

Matplotlib/Seaborn Plotting Engine

Generates static statistical and relational plots.

HoloViews/hvPlot Plotting Engine

Provides an interactive visualization backend for dynamic plots.

Visualization Output & Export Manager

Manages saving and displaying all generated visualizations.