Sketch is an open-source AI-powered data analysis assistant designed specifically for pandas users, enabling natural language interaction with tabular datasets to generate code, insights, and transformations. It works by summarizing the structure and statistical properties of a dataset and providing that context to a language model, allowing it to generate highly relevant and accurate responses tailored to the data. The tool integrates directly into pandas dataframes through an extension, making it easy to use within existing Python workflows without requiring additional IDE plugins. Sketch supports a variety of tasks including data cleaning, feature engineering, visualization, and exploratory analysis, all driven by simple natural language prompts. It also includes advanced capabilities for generating structured outputs and applying transformations directly to datasets, reducing the need for manual coding.
Features
- Natural language interface for pandas dataframes
- Automatic dataset summarization for context aware responses
- Code generation for data cleaning analysis and visualization
- Direct dataframe integration without IDE plugins
- Support for feature engineering and transformation workflows
- Lightweight design using efficient data approximation methods