StatSC (USC)

Statistics-Sensometry-Chimiometry

The integration of multi-source data is more topical than ever, as the data collected is increasingly complex and its volume continues to grow due to the development of analytical platforms, imaging techniques, the boom in omics data, the introduction of new methods for collecting sensory data, closer to the consumer... Against this backdrop, StatSC's scientific project aims to provide methodological solutions through the development of specific methods for the joint analysis of several data tables (structured, multi-block, multi-channel data).

The developments carried out by StatSC are both exploratory (dimensionality reduction, classification) and predictive (regression and discrimination). At the same time, more conceptual work is being carried out to bring the various methods closer together, so as to place them within a unified formal framework, clarify them and assess their respective merits in meeting various objectives. This methodological foundation contributes to providing analysis tools in a wide range of application fields, with a view to integrating or merging multi-source data.

• Unsupervised multi-block approach: Dimensionality reduction and Classification.
• Supervised multi-block approach: Discrimination and Prediction.
• Sensory and preference data, -omics data, hyperspectral image data

Unit Manager :

Evelyne VIGNEAU

Deputy Director :

Mohamed HANAFI