Getting started

pandasx brings pandas' Series and DataFrame to Node.js. If you know pandas, you already know most of pandasx — names are camelCase and a few Python-only syntaxes become methods (see Coming from pandas).

Installation

npm install pandasx

pandasx needs Node.js 22.23.4+, 24.22.1+ or 26.12.1+ (earlier releases have the Node2Shell Brotli bug) and has no required dependencies. Some file formats use optional packages that you install only if you need them:

Format	Install
Excel (.xlsx)	npm install exceljs
Parquet	npm install hyparquet hyparquet-writer
Feather / Arrow	npm install apache-arrow
HDF5	npm install h5wasm

If you call a reader whose package is missing, the error tells you exactly what to install.

Your first DataFrame

import * as pd from 'pandasx';

const df = new pd.DataFrame({
  city: ['São Paulo', 'Rio', 'Recife', 'Curitiba'],
  population: [12.4, 6.7, 1.6, 1.9],
  coastal: [false, true, true, false],
});

console.log(df);
console.log(df.dtypes);

        city  population  coastal
0  São Paulo        12.4    false
1        Rio         6.7     true
2     Recife         1.6     true
3   Curitiba         1.9    false
city         string
population  float64
coastal        bool
dtype: string

console.log prints frames the way pandas does. Columns are typed: population is stored in a Float64Array, coastal in a Uint8Array.

A typical analysis

import * as pd from 'pandasx';

const sales = new pd.DataFrame({
  region: ['North', 'South', 'North', 'East', 'South', 'North'],
  product: ['coffee', 'coffee', 'tea', 'coffee', 'tea', 'coffee'],
  quantity: [10, 4, 0, 7, 12, 25],
  price: [12.5, 13, 8, 11.9, 7.5, 12.5],
});

const summary = sales
  .query('quantity > 0')
  .assign({ revenue: (d) => d.get('quantity').mul(d.get('price')) })
  .groupby('region')
  .agg({ revenue: 'sum', quantity: 'mean' })
  .sortValues('revenue', { ascending: false });

console.log(summary);

        revenue  quantity
region
North     437.5      17.5
South     142.0       8.0
East       83.3       7.0

Reading files

Readers mirror pandas' read_* functions:

const df = pd.readCsv('sales.csv', { parseDates: ['date'], indexCol: 'id' });
const big = pd.readCsvChunks('huge.csv.gz', { chunksize: 100_000 }); // async iterator
const xlsx = await pd.readExcel('report.xlsx', { sheetName: 'data' });

See Reading & writing data for every format.

TypeScript

pandasx ships its own type declarations. Methods return precise types (Series, DataFrame, DataFrameGroupBy, …) and options objects are fully typed, so your editor can autocomplete everything. Values inside a Series are typed as unknown, because a column's dtype is only known at run time — cast when you need a specific type:

const total = df.get('population').sum();        // number
const first = df.get('city').iat(0) as string;   // unknown → string

Next steps

• Data structures & dtypes: Series, DataFrame, Index and every dtype.
• Selecting & assigning: get, loc, iloc, query and in-place assignment.
• Group by, Reshaping and Time series for analysis.
• The API reference documents every method and option.
• The examples folder has 20 runnable scripts, one per feature area, on a shared dataset.