Node.js and Go Alternatives for Pandas and NumPy in 2026: 8 Tools Compared

Danfo.js is the first library to try if you want pandas in JavaScript. You can load data from arrays, JSON, CSV or tensors, create a DataFrame, filter rows, add columns, handle missing values, group and aggregate, merge tables and plot results. The API deliberately mirrors pandas, so a Python user can often guess method names and get productive quickly.

It is built on TensorFlow.js for low-level maths, which also means you can convert a DataFrame to a tensor for machine learning. The limits are ecosystem and scale: Danfo.js does not have the huge library of Python data tools around pandas, and very large tables may need chunking or a server-side approach. For web apps, dashboards and Node.js services that need pandas-style tables without Python, Danfo.js is the closest fit.

TensorFlow.js is the NumPy-adjacent library for JavaScript machine learning. It provides tensors, array maths, automatic differentiation and model training and inference, in the browser and in Node.js. If your pandas and NumPy work ends in a model, TensorFlow.js lets you keep that whole pipeline in JavaScript, including running a trained model in a web app without a Python server.

It does not provide pandas-style labelled tables, so pair it with Danfo.js for data preparation. A NumPy user will recognise many tensor concepts, but method names and memory management differ, and you need to dispose tensors carefully in long-running apps. For JavaScript teams that need numerical arrays and machine learning, TensorFlow.js is the standard choice.

math.js is a general maths library for JavaScript, covering numbers, big numbers, complex numbers, fractions, units, matrices and an expression parser. If your NumPy use is mainly matrix maths, statistics and calculations in a web app, math.js gives you a clear, well-documented API without the weight of a machine learning framework. It also handles units, which helps in science and engineering apps.

It is not a replacement for pandas, and it is not optimised for very large arrays in the way NumPy is. Use math.js for calculations, formulas and smaller matrices, and use Danfo.js or TensorFlow.js when you need tables or tensors at scale. For many JavaScript apps, math.js plus Danfo.js covers the practical ground that pandas and NumPy cover in Python.

Arquero, from the University of Washington Interactive Data Lab, is a JavaScript library for query processing and transformation of array-backed tables. It offers verbs for selecting, filtering, deriving, grouping, joining and rolling up data, with a design influenced by dplyr and SQL as much as pandas. It is fast for in-browser data work and pairs well with Arrow-based data formats.

A pandas user will need to learn Arquero verbs, because method names and chaining differ from pandas. That learning pays off in data-heavy web apps, where Arquero handles large tables smoothly and composes well with visualisation tools. If you want a pandas-like API above all, Danfo.js fits better. If you want fast, expressive table transforms in a JavaScript app, Arquero is a strong alternative.

Data-Forge is a JavaScript and TypeScript library for working with tabular and time-series data. It provides DataFrame and Series types, with methods for parsing, filtering, transforming, grouping and aggregating data, plus tools for working with dates and series. TypeScript users will appreciate its typed API, which catches errors earlier than plain JavaScript.

Its community is smaller than Danfo.js, and development has been slower, so check the current release and docs before building a new project on it. For a TypeScript team that wants a clear, LINQ-style data wrangling API and does not need machine learning integration, Data-Forge remains a usable pandas alternative. For a new project in 2026, Danfo.js is usually the safer first choice.

Gonum is the closest Go equivalent to NumPy for numerical work. It provides packages for matrices, linear algebra, statistics, optimisation, spatial data and more, designed for Go developers who need reliable numerical code in a service or tool. If you are building a Go backend that does calculations, simulations or data analysis, Gonum gives you the building blocks NumPy gives Python developers.

Gonum does not include a pandas-style DataFrame, so pair it with DataFrame-go for tabular work, or use plain Go slices and maps for simple cases. Go code will be more verbose than NumPy for array operations, because Go lacks NumPy broadcasting and vectorised syntax. That verbosity buys you compiled speed, static typing and easy deployment as a single binary, which suits many production services.

Gorgonia is a Go library for building machine learning models as expression graphs, with automatic differentiation similar to early TensorFlow. It lets Go developers define a computation graph, run it and train models, without calling out to Python. That matters for teams that want a single-language Go service, from data ingestion to model inference, with no Python runtime to manage.

The learning curve is steeper than a high-level framework, because you work at the graph level rather than calling a ready-made model class. The ecosystem is also smaller than Python machine learning, so you will write more code for data loading and evaluation. For Go teams doing custom models or research, Gorgonia is the most direct NumPy and TensorFlow-style option in Go.

DataFrame-go provides a DataFrame type for Go, aimed at statistics, machine learning and data manipulation. You can load data, select columns, filter rows, group and aggregate, and join tables, in a style familiar to pandas users. That makes it a natural partner to Gonum: DataFrame-go handles the table, Gonum handles the maths.

Its feature set is smaller than pandas, and its maintenance pace is slower than larger Go projects, so check the repository, issues and latest release before you depend on it. For a Go service that needs modest tabular work, DataFrame-go saves a lot of manual slice and map code. For heavy data analysis, a Python service or a JavaScript Danfo.js workflow may still be more productive.