EN • ES

Fit probability distributions online with Phitter

BETA

Phitter analyzes datasets and determines the best analytical probability distributions that represent them. The Phitter kernel studies over 80 probability distributions, both continuous and discrete, 3 goodness-of-fit tests, and interactive visualizations. For each selected probability distribution, a standard modeling guide is provided along with spreadsheets that detail the methodology for using the chosen distribution in data science, operations research, and artificial intelligence.

Phitter is exhaustive and free

Get StartedDistributions

From observations to a distribution.

Phitter compares analytical probability distributions against your sample and ranks the candidates using the sum of squared errors.

For each fitted model, it reports the estimated parameters, goodness-of-fit results, and the plots needed to inspect the match. The calculation runs in Python through Pyodide, directly in your browser.

Fit a dataset Browse distributions Read the Phitter blog Watch the Galton board

One analysis. Every result in view.

Paste observations or import a TXT, CSV, XLSX, or remote data file. Choose continuous or discrete models, run the fit, and examine the evidence before selecting a distribution.

Input
TXT · CSV · XLSX · pasted values
Your sample stays in the browser.
Estimation
Parameter fitting and model ranking
Candidates are ordered by sum of squared errors.
Diagnostics
Histogram or PMF · PDF · CDF · Q–Q
Visual checks accompany the numerical results.

Eighty-seven distributions. Three tests.

The same interface covers continuous and discrete data without reducing the analysis to a single score.

Continuous
77 distributions
Density, cumulative function, quantiles, moments, and fitted parameters.
Discrete
10 distributions
Probability mass, cumulative function, quantiles, moments, and fitted parameters.
Goodness of fit
KS · AD · χ²
Kolmogorov–Smirnov, Anderson–Darling, and chi-square where applicable.

A ranking is evidence, not a verdict.

The first model in a ranking is a candidate. Sample size, the process that generated the observations, parameter uncertainty, test assumptions, and visual diagnostics still matter.

Phitter keeps those details visible so the final choice remains a statistical decision rather than an automatic label.

Local by design. Open by construction.

Phitter does not need an account and does not send your observations to a Phitter fitting server. The open-source Python kernel runs locally through Pyodide.

The web application is free to use, and the kernel, documentation, formulas, and implementation are publicly available for inspection.

A public scientific record.

Phitter has a peer-reviewed software paper, and its open-source kernel is publicly available on GitHub.