Welcome to the Carotta Home Page!

What is Carotta?

Carotta is a software application that provides easy access to advanced unsupervised learning analysis specifically designed for breath data analysis.

Short Summary

We are addressing two main goals, a user-friendly front-end including several visualization options as well as a flexible and modular back-end that is open for functional extensions. Carotta guides the user through the different steps of unsupervised learning analysis, starting with the similarity function, clustering, cluster quality evaluation, filtering, and visualization by dimension reduction. Thereby, it offers biomedical researchers access to this techniques without requiring deeper knowledge on advanced learning techniques. The software application provides an intuitive way to process and analyze their data efficiently, grasping back to well established machine learning technologies in the background. The flexible plug-in system allows future methods to be added in a straight-forward fashion. Each step comes with an interactive visualization allowing for in-depth investigation of intermediate results directly in the user interface without the necessity to install and configure external software packages or libraries.

Where is Carotta developed?

Carotta was developed at the Max Planck Institute Informatik, Saarbr├╝cken, in cooperation with the University of Southern Denmark.


A.C. Hauschild, Max Planck Institute for Informatics, Saarbr├╝cken, Germany and University of Southern Denmark
Tobias Frisch, Department of Mathematics and Computer Science, Freie Universitaet Berlin, Berlin; Germany


Carotta - Revealing Hidden Confounder Markers in Metabolic Breath Profiles.
A.-C. Hauschild, T. Frisch, J.I. Baumbach, J. Baumbach