Running in Pyodide

The actual Python skaters package, compiled to WebAssembly and executed in your browser — no server, no install. Pure Python, zero dependencies.

The JavaScript port on the playground is fast and offline. This page instead runs the real Python package via Pyodide — proof that the same library you pip install runs unmodified in the browser. The two are verified to agree to 1e-6 by the parity test suite.

Pyodide is ~10 MB; the first run downloads it.

observation forecast mean ±2σ

The code that runs

import micropip
await micropip.install("skaters")

import random, math
from skaters import laplace

f = laplace(k=1)
state = None
for y in series:           # one observation at a time
    dists, state = f(y, state)
    mean = dists[0].mean   # point forecast
    std  = dists[0].std    # uncertainty