Data fusion using Multiple Factor Analysis coupled with non-linear pattern recognition (fuzzy k-means): application to Chenin blanc

Patterns in data obtained from wine chemical and sensory evaluations are difficult to decipher using classical statistics. Coupling data fusion with machine learning techniques could assist in solving these issues and lead to new hypotheses. The current study investigated the applicability of classical and machine learning pattern recognition approaches for oenological applications. A sample set of 23 Chenin blanc wines made from young (< 35 years) and old (> 35 years) vines were analysed (recently bottled (Year 1) and after two years of storage (Year 2)). Sensory (sorting) and chemical (NMR: nuclear magnetic resonance and HRMS: high-resolution mass spectrometry) data were collected. Multiple factor analysis (MFA) was used for the data fusion. Cluster analysis was performed by agglomerative hierarchical clustering (AHC) and fuzzy k-means. Optimal cluster conditions were found for both methods and the cophenetic coefficient was used to assess the confidence of fit. Given the large number of variables, the models were complex. Inconsistent clustering patterns were observed when varying clustering conditions, indicating high similarity between samples. Overall, fuzzy k-means resolved clustering patterns better than AHC and, coupled with data fusion, improved the interpretation of the complex oenological data.

Abstract licence: CC BY

34 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.