PollisBright Insights through Statistical Learning, Simulation and Optimisation

Had an idea? Pollis helps you test it, in code you own, and then improve and expand it.

A free, open-source desktop environment to explore data, build models, run simulations and solve optimisation problems: from your first click to your own code in Julia.

Download for macOSApple silicon (M1 and later)
Installing on macOS
  1. Download Pollis for Mac (Pollis-darwin-arm64.zip).
  2. Unzip it (double-click it in Downloads).
  3. Move Pollis.app to Applications (drag it onto Applications in the Finder sidebar).
  4. Then in Terminal (press Command-Space, type Terminal, press Return) run xattr -cr /Applications/Pollis.app and press Return (the app is not signed yet).
  5. Open it by double-clicking Pollis.app in Applications.
Download for Windows64-bit
Installing on Windows
  1. Download Pollis for Windows (Pollis-win32-x64.zip).
  2. Unzip it (right-click it in Downloads, Extract All..., Extract).
  3. Open the folder Pollis-win32-x64.
  4. Open it by double-clicking Pollis.exe. If SmartScreen warns, choose More info, then Run anyway.
Download for Linuxx64, .tar.gz
Installing on Linux
  1. Download Pollis for Linux (Pollis-linux-x64.tar.gz).
  2. Extract it in a terminal with tar -xzf Pollis-linux-x64.tar.gz.
  3. Run Pollis-linux-x64/pollis.

Pollis is young and growing quickly. Some features are still being built, and you may meet the odd rough edge. Your feedback at [email protected] helps decide what comes next.

Pollis with the Model menu open, listing distributions, regression, time series, survival analysis, Bayesian analysis and machine learning methods, and the three Pollis Toolboxes in the Extensions view

Data and a question? Start here.

Pick a method from a menu, fill in a few boxes, and Pollis gives you working Julia code that produces the result, with advice on what a careful analyst does next. You do not need to program to start.

Explore

What do my data look like?

Data from finance, economics and AI libraries, Excel, CSV or SQL. Thirty-five plot types, descriptive statistics, outliers and missing data.

Model

What explains it?

Regression with robust inference, generalised linear models, time series, survival analysis, Bayesian inference, machine learning and text analysis.

Simulate

What could happen?

Monte Carlo and Markov chain Monte Carlo, bootstrap, stochastic differential equations, agent-based and network models.

Optimise

What is the best decision?

Linear, convex and nonlinear programming, global and Bayesian optimisation, optimal control and reinforcement learning.

Compose

How do I put it all together?

Workflows that replicate classic papers, reports, learning activities, new methods as packages, collaboration, coding agents with their skills, and local reasoning models.

Everything in Pollis

5 curated menus · 204 method panels · 657 procedures · 289 open-source Julia packages. Items marked › open a submenu; one is shown open.

All five Pollis menus with their entries: Explore, Model, Simulate, Optimise and Compose, with the Reinforcement Learning submenu of Optimise shown open

Inside a panel

Each of the 204 panels brings together one or more methods and their procedures, with everything you need around them.

The Generalised Linear Models panel: Powered by, Key Points, Decision Table, Example Code with family tabs, link and inference choices, the code, its action buttons, Next Steps and Learn More 1 2 3 4 5 6 7 8 9
The Generalised Linear Models panel. Choosing a family, a link or a kind of inference (Standard, HC0–HC3, HAC or Clustered) rewrites the example code.
  1. 1Powered by. The open-source packages behind the method, with links. A tick when they are installed, or a one-click Install.
  2. 2Key Points. The method in a few lines.
  3. 3Decision Table. Which variant to use, and when.
  4. 4Your choices. Family, link function and kind of inference: tabs and drop-down lists.
  5. 5Your own tab. Click + to add a tab of your own, with your own code.
  6. 6Example code. Real Julia code that follows your choices, with comments.
  7. 7Run it your way. Edit or copy the code, or send it directly to the Julia REPL, the Editor, or a Jupyter or Pluto notebook.
  8. 8Next Steps. Ready-made code to Visualise, Diagnose, Predict, Compare and Interpret the fitted model.
  9. 9Learn More. Local wikis, notebook tutorials, video tutorials and references.

Why Julia

Julia is a modern, open-source language for scientific computing: as easy to read as Python or R, and as fast as C or Fortran. Its packages cover statistics, machine learning, differential equations and optimisation, and Pollis builds on them, so what you learn in Pollis carries over to Julia itself.

Free, open and yours

No licence fees, no lock-in, and code you can take anywhere.

Keep the code

Every panel writes readable Julia code that is yours, ready for the REPL, an editor or a notebook.

Learn good practice

Which method to choose, how to check it, and where to read more.

A familiar editor

Built on Code - OSS, so it looks and works like a code editor many people already know.

Extensions

Add Pollis Toolboxes for whole fields, and thousands of editor extensions from Open VSX: programming languages, remote development, cloud integrations, code quality, productivity tools, themes and more.

Open source

Licensed under the AGPL 3.0, on top of open-source Julia packages. Windows, macOS and Linux.

Private by default

Pollis runs on your own computer. Your data stays with you, and AI features are off unless you turn them on.

Learn more

A Tour of PollisTwelve pages: what Pollis is, inside a panel, coverage, customisation, reports, practice and collaboration.Download PDF · 1.5 MB

The name

Pollis echoes the Greek polis, the community of citizens and the rules it freely adopts. Like a polis, Pollis is shared: free, open source, built on open-source packages, and open to change by those who use it. Its double l recalls the Greek pollá, “many things”: many fields and many methods in one place.