Pandas for JavaScript, with TensorFlow hooks

What it does

Danfo.js is a TypeScript library that brings labeled data structures—DataFrame and Series—to JavaScript and Node.js. It mimics the Pandas API closely so Python data practitioners can transfer their mental models without friction. You can ingest CSV, JSON, or Excel files, clean and transform tabular data, and convert structures directly into TensorFlow.js tensors.

The interesting bit

The library runs in both Node.js and the browser, bundling interactive plotting and machine-learning preprocessing into a single toolkit. Its tight integration with TensorFlow.js is the unusual angle: it treats dataframes not as dead-end spreadsheets, but as staging areas for ML pipelines in pure JavaScript.

Key highlights

• Pandas-like API with DataFrame, Series, groupby, merge, and join
• Native TensorFlow.js tensor conversion for ML workflows
• Built-in interactive plotting for DataFrames and Series in the browser
• IO tools for CSV, JSON, and Excel ingestion
• Data preprocessing utilities including StandardScaler, MinMaxScaler, and encoders
• Notebook support for VS Code and ObservableHQ

Verdict

Worth a look if you are doing data exploration or ML preprocessing in JavaScript and want to avoid context-switching to Python. Pure Python data scientists already happy with Pandas and PyPlot will find it redundant.