Skip to content

formulaML reference docs

This site is the reference manual for the ML.* custom functions in the formulaML Excel add-in. Every function you can call from a cell — ML.CLASSIFICATION.LOGISTIC, ML.FIT, ML.PREDICT, ML.PIPELINE, and the rest — has its own page here listing its arguments, its return type, and at least one working Excel example.

Where to start

  • New to formulaML? Read Getting started to install the add-in and train your first model.
  • Curious how a cell can hold a Python object? See the Object handles concept page.
  • Building your first model? Read The FIT/PREDICT pattern — every classifier, regressor, and clusterer follows the same three-step workflow.
  • Looking up a specific function? Jump straight to the Reference — namespaces are listed alphabetically.

Drive formulaML with Claude

Want to drive formulaML from natural language directly inside Excel? Install the formulaML Claude skill in your Claude account and the Claude add-in for Excel will pick it up automatically — letting you ask Claude to load datasets, train models, score them, and predict, all in plain English. The skill teaches Claude how to chain ML.* formulas, lay out cells, and translate model output into plain English.

Download the formulaML Claude skill (zip)

To install:

  1. Download the zip above and unzip it — you should get a formulaML/ folder containing six .md files.
  2. In Claude Desktop or on claude.ai, open your skills settings and upload the unzipped formulaML/ folder as a new skill.
  3. Open Excel, launch the Claude add-in, and ask it anything about your data — for example, "predict the target column using the data in Sheet1!A1:F200". Claude will use the skill to drive the ML.* formulas for you.

The Claude add-in for Excel reads skills directly from your Claude account, so installing the skill once via Desktop or web makes it available in Excel without any further setup.