Khiops – An Open-Source Library for Structured Data Analysis

Speakers
-
Alexis BonduResearcher in Machine Learning, Orange Innovation
-
Luc-Aurélien GauthierData Scientist and Head of Khiops Product, Orange
Webinar Abstract
Our Speakers: Alexis and Luc-Aurélien, representatives of Orange, will discuss the following topic:
Khiops – An Open-Source Library Specialized in Structured Data Analysis
- Developed and tested at Orange for 20 years, Khiops is an open-source library specialized in structured (tabular) data analysis.
- It stands out due to a unique formalism with particularly innovative properties (derived from over a hundred research papers), such as automatic and robust handling of relational data (auto feature engineering on multi-table data).
- Prediction models are inherently interpretable, allowing users to understand the factors influencing predictions. Additionally, the learning process is efficient, fast, and highly energy-efficient.
- Khiops has been successfully applied to numerous real-world use cases at Orange, enabling significant productivity gains in production services.
- This recent transition to Open Source (under the BSD-3-Clause-Clear license) aligns with Orange’s commitment to sharing its expertise in AI and fostering collaboration within the scientific community.