v3.0.0-beta.0

What's Changed

Features

• VirtualDataFrame: Introduced virtual dataframes for on-demand data loading and direct_sql functionality. ([#1434](#1434))
• Dataframe Enhancements:
  • Save dataframe as YAML. ([#1443](#1443))
  • Push and pull datasets to/from remote servers. ([#1445](#1446))
  • Added get_default_schema for default schema generation. ([#1525](#1525))
• SemanticLayerSchema: New schema class for logical and business validation, with relationship validation and view support. ([#1516](#1534))
• Global Config: Unified configuration system across the application. ([#1467](#1467))
• BambooLLM: Updated implementation for the BambooLLM. ([#1462](#1462))

Refactors

• Removed deprecated connectors, pipelines, and agents for a cleaner workflow. ([#1433](#1433), [#1435](#1435))
• Replaced environment variables PANDASAI_* with PANDABI_* for clarity. ([#1511](#1511))
• Simplified dataframe serialization and default naming. ([#1523](#1523))

Fixes

• Resolved dependency issues causing build failures. ([#1508](#1508))
• Addressed bugs in dataloader for CSV and Parquet files. ([#1451](#1451))
• Fixed virtual dataframe naming conflicts. ([#1531](#1531))

Documentation

• Reorganized and updated docs for v2 and v3, including new notebooks and examples. ([#1505](#1505), [#1522](#1526))
• Added a new GIF in the README for v3. ([#1536](#1536))

New Contributors

• @gdcsinaptik, @scaliseraoul, and @ellipsis-dev joined with impactful contributions.

Full Changelog: [View on GitHub](v3.0.0a11...v3.0.0-beta.0)