Return
Integrating multivariate statistics and interpretable machine learning for the quantitative profiling of tea polyphenols and catechins in six tea categories
J
Q
Y
L
Y
G
Y
X
H
DOI:10.1016/j.foodres.2026.120295.png)
Abstract
En 中文
• A large-scale analysis of 1381 tea samples reveals tea polyphenols and catechins transformation across six tea types. • Ester-type catechins show a distinct gradient decrease from green to dark tea. • The SMOTE-L1-XGBoost model achieves high metrics with Macro F1 = 0.883 and AUC = 0.945. • SHAP identified EGCG as the dominant phenotypic contributor in green and black tea.
Keywords:
Tea polyphenols
Catechins
Processing technology
Machine learning
SHAP analysis
GT
,
green tea
WT
,
white tea
YT
,
yellow tea
OT
,
oolong tea
BT
,
black tea
DT
,
dark tea
ETC
,
ester-type catechins
NETC
,
non-ester-type catechins
EGCG
,
(−)-epigallocatechin gallate
ECG
,
(−)-epicatechin gallate
EGC
,
(−)-epigallocatechin
EC
,
(−)-epicatechin
C
,
(+)-catechin
TP
,
tea polyphenols
TC
,
total catechins
GA
,
gallic acid
ETC/NETC
,
the ratio of esterified to non-esterified catechins
EGCG/TC
,
the ratio of epigallocatechin gallate to total catechins
EGCG/TP
,
the ratio of epigallocatechin gallate to tea polyphenols
TC/TP
,
the ratio of total catechins to tea polyphenols
PPO
,
polyphenol oxidase
POD
,
peroxidase
TFs
,
theaflavins
TRs
,
thearubigins
TBs
,
theabrownins
VIP
,
variable importance in projection
ROC
,
receiver operating characteristic
AUC
,
area under the curve value
RF
,
Random Forest
XGBoost
,
eXtreme Gradient Boosting
LightGBM
,
Light Gradient Boosting Machine
PCA
,
principal component analysis
PLS-DA
,
partial least squares discriminant analysis
ADASYN
,
Adaptive Synthetic Sampling
SMOTE
,
Synthetic Minority Oversampling Technique
MI
,
mutual information
L1
,
L1 regularization
RFECV
,
recursive feature elimination with cross-validation
SHAP
,
SHapley Additive exPlanations
Journal
IF:
8
Papers:
1.7W
Citations:
7.9W
