Table 2 Selection of hyperparameters for the DT, RF, KNN, SVM, GB, ET, AB and ANN models.
Models | Selected hyperparameter for the men model to predict healthy, osteopenia, and OP | Selected hyperparameter for the women model to predict healthy, osteopenia, and OP | Selected hyperparameter for the men model to propose sports protocols | Selected hyperparameter for the women model to propose sports protocols |
|---|---|---|---|---|
DT | max_depth = 4, max_leaf_nodes = 6 | max_depth = 5, max_leaf_nodes = 7 | max_depth = 3, max_leaf_nodes = 7 | max_depth = 6, max_leaf_nodes = 12 |
RF | max_depth = 7, random_state = 3, n_estimators = 19 | max_depth = 10, random_state = 17, n_estimators = 9 | max_depth = 7, random_state = 14, n_estimators = 13 | max_depth = 13, random_state = 7, n_estimators = 16 |
KNN | n_neighbors = 3 | n_neighbors = 10 | n_neighbors = 2 | n_neighbors = 4 |
SVM | C = 17, kernel = 'rbf', gamma = 'scale', coef0 = 0.0, shrinking = True, probability = True, tol = 0.001, class_weight = None, verbose = False, decision_function_shape = 'ovr', break_ties = False, random_state = 0 | C = 28, kernel = 'rbf', gamma = 'scale', coef0 = 0.0, shrinking = True, probability = True, tol = 0.001, class_weight = None, verbose = False, decision_function_shape = 'ovr', break_ties = False, random_state = 0 | C = 15, kernel = 'rbf', gamma = 'scale', coef0 = 0.0, shrinking = True, probability = True, tol = 0.001, class_weight = None, verbose = False, decision_function_shape = 'ovr', break_ties = False, random_state = 0 | C = 37, kernel = 'rbf', gamma = 'scale', coef0 = 0.0, shrinking = True, probability = True, tol = 0.001, class_weight = None, verbose = False, decision_function_shape = 'ovr', break_ties = False, random_state = 0 |
GB | max_depth = 22, min_samples_split = 8, min_samples_leaf = 14, max_features = 2 | max_depth = 3, min_samples_split = 9, min_samples_leaf = 6, max_features = 5 | max_depth = 13, min_samples_split = 14, min_samples_leaf = 12 | max_depth = 6, min_samples_split = 10, min_samples_leaf = 7 |
ET | max_depth = 8, min_samples_split = 9, min_samples_leaf = 3, max_features = 11 | max_depth = 15, min_samples_split = 3, min_samples_leaf = 6, max_features = 15 | max_depth = 11, min_samples_split = 2, min_samples_leaf = 1, max_features = 8 | max_depth = 22, min_samples_split = 2, min_samples_leaf = 2, max_features = 15 |
AB | n_estimators = 10, base_estimator = base, learning_rate = 0.1 | n_estimators = 4, base_estimator = base, learning_rate = 0.1 | n_estimators = 29, base_estimator = base, learning_rate = 0.1 | n_estimators = 13, base_estimator = base, learning_rate = 0.1 |
ANN | activation = 'tanh', alpha = 0.0001, batch_size = 'auto', hidden_layer_sizes = 35, learning_rate = 'constant', max_iter = 470 | activation = 'tanh', alpha = 0.0001, batch_size = 'auto', hidden_layer_sizes = 25, learning_rate = 'constant', max_iter = 754 | activation = 'tanh', alpha = 0.0001, batch_size = 'auto', hidden_layer_sizes = 30, learning_rate = 'constant', max_iter = 470 | activation = 'tanh', alpha = 0.0001, batch_size = 'auto', hidden_layer_sizes = 27, learning_rate = 'constant', max_iter = 754 |
