diff --git a/results/best_results_accuracy_AdaBoostStree.json b/results/best_results_accuracy_AdaBoostStree.json new file mode 100644 index 0000000..e8dd383 --- /dev/null +++ b/results/best_results_accuracy_AdaBoostStree.json @@ -0,0 +1,491 @@ +{ + "balance-scale": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 10000.0, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "balloons": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "breast-cancer-wisc-diag": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.2, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "breast-cancer-wisc-prog": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.2, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "breast-cancer-wisc": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "rbf", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "breast-cancer": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "cardiotocography-10clases": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "cardiotocography-3clases": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "conn-bench-sonar-mines-rocks": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "rbf", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "cylinder-bands": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "rbf", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "dermatology": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 55, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "echocardiogram": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__splitter": "random", + "base_estimator__max_features": "auto" + }, + "-program made-" + ], + "fertility": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.05, + "base_estimator__max_features": "auto", + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "haberman-survival": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "heart-hungarian": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.05, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "hepatitis": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "ilpd-indian-liver": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "ionosphere": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "iris": [0.0, { "n_estimators": 100, "n_jobs": -1 }, "-program made-"], + "led-display": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "libras": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.08, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "low-res-spect": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "lymphography": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.05, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "mammographic": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "rbf", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "molec-biol-promoter": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "rbf", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "musk-1": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.05, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "poly", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "oocytes_merluccius_nucleus_4d": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 8.25, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "poly", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "oocytes_merluccius_states_2f": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "oocytes_trisopterus_nucleus_2f": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "oocytes_trisopterus_states_5b": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.11, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "parkinsons": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "pima": [0.0, { "n_estimators": 100, "n_jobs": -1 }, "-program made-"], + "pittsburg-bridges-MATERIAL": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "pittsburg-bridges-REL-L": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "pittsburg-bridges-SPAN": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.05, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "pittsburg-bridges-T-OR-D": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__splitter": "random", + "base_estimator__max_features": "auto" + }, + "-program made-" + ], + "planning": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__gamma": 10.0, + "base_estimator__kernel": "rbf", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "post-operative": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 55, + "base_estimator__degree": 5, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "poly", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "seeds": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 10000.0, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "statlog-australian-credit": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "statlog-german-credit": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "statlog-heart": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "statlog-image": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 7, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "statlog-vehicle": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "synthetic-control": [ + 0.0, + { "n_estimators": 100, "n_jobs": -1 }, + "-program made-" + ], + "tic-tac-toe": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.2, + "base_estimator__gamma": 0.1, + "base_estimator__kernel": "poly", + "base_estimator__max_iter": 10000.0, + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "vertebral-column-2clases": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "wine": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.55, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ], + "zoo": [ + 0.0, + { + "n_estimators": 100, + "algorithm": "SAMME", + "base_estimator__C": 0.1, + "base_estimator__max_iter": 10000.0, + "base_estimator__kernel": "liblinear", + "base_estimator__multiclass_strategy": "ovr" + }, + "-program made-" + ] +} diff --git a/src/Models.py b/src/Models.py index 5b12155..ef5ab72 100644 --- a/src/Models.py +++ b/src/Models.py @@ -1,6 +1,10 @@ from statistics import mean from sklearn.tree import DecisionTreeClassifier, ExtraTreeClassifier -from sklearn.ensemble import RandomForestClassifier, BaggingClassifier +from sklearn.ensemble import ( + RandomForestClassifier, + BaggingClassifier, + AdaBoostClassifier, +) from sklearn.svm import SVC from stree import Stree from wodt import Wodt @@ -28,6 +32,9 @@ class Models: if name == "BaggingWodt": clf = Wodt(random_state=random_state) return BaggingClassifier(base_estimator=clf) + if name == "AdaBoostStree": + clf = Stree(random_state=random_state) + return AdaBoostClassifier(base_estimator=clf) if name == "RandomForest": return RandomForestClassifier() msg = f"No model recognized {name}" @@ -47,7 +54,7 @@ class Models: nodes = 0 leaves = result.get_n_leaves() depth = 0 - elif name.startswith("Bagging"): + elif name.startswith("Bagging") or name.startswith("AdaBoost"): if hasattr(result.base_estimator_, "nodes_leaves"): nodes, leaves = list( zip(*[x.nodes_leaves() for x in result.estimators_])