Skip to main content
. 2025 Mar 25;10(13):13502–13514. doi: 10.1021/acsomega.5c00075

Table 2. Result of the Different ML Models.

model best parameters accuracy (%) precision (%)
LR ‘solver’: ‘liblinear’, ‘penalty’: ‘l2’ 66.38 68.31
k-NN ‘weights’: ‘distance’, ‘n_neighbors’: 13, ‘leaf_size’: 1, ‘algorithm’: ‘kd_tree’ 59.48 63.93
NB ‘priors’: None 65.09 66.33
RF ‘n_estimators’: 200, ‘min_samples_split’: 10, ‘min_samples_leaf’: 2, ‘max_features’: ‘sqrt’, ‘max_depth’: 20, ‘criterion’: ‘gini’, ‘bootstrap’: False 71.98 72.22
DT ‘splitter’: ‘random’, ‘min_samples_split’: 10, ‘min_samples_leaf’: 10, ‘max_features’: ‘log2’, ‘max_depth’: 3, ‘criterion’: ‘entropy’ 62.93 62.77
QDA ‘reg_param’: 0.1, ‘priors’: None 65.95 66.67
MLP classifier ‘solver’: ‘sgd’, ‘max_iter’: 500, ‘learning_rate’: ‘adaptive’, ‘hidden_layer_sizes’: (50,), ‘alpha’: 0.001, ‘activation’: ‘logistic’ 58.19 62.77