""" =================== 2. ML Experiments =================== """ import matplotlib.pyplot as plt from SeqMetrics import ClassificationMetrics from ai4water.experiments import MLClassificationExperiments from utils import SAVE from utils import version_info, return_train_test, set_rcParams # %% version_info() # %% set_rcParams() # %% def f1_weighted(t,p)->float: return ClassificationMetrics(t, p).f1_score(average="weighted") def f1_micro(t,p)->float: return ClassificationMetrics(t, p).f1_score(average="micro") def precision_weighted(t,p)->float: return ClassificationMetrics(t, p).precision(average="weighted") def precision_micro(t,p)->float: return ClassificationMetrics(t, p).precision(average="micro") def recall_weighted(t,p)->float: return ClassificationMetrics(t, p).recall(average="weighted") def recall_micro(t,p)->float: return ClassificationMetrics(t, p).recall(average="micro") # %% # WQ and Antibiotics # ================== scenario = 'no_genes' # %% # CTX-M # ------ target = 'CTX-M' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # OXA48 # ------- target = 'OXA48' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # TEM # ------ target = 'TEM' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # MCR-1 # ------ target = 'MCR-1' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8),) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # Only WQ # ========= scenario = 'only_wq' # %% # CTX-M # ------ target = 'CTX-M' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors('f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # OXA48 # ------- target = 'OXA48' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # TEM # ------ target = 'TEM' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # MCR-1 # ------ target = 'MCR-1' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # Only Antibiotics # ================ scenario = 'only_antibiotics' # %% # CTX-M # ------ target = 'CTX-M' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors('f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # OXA48 # ------- target = 'OXA48' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # TEM # ------ target = 'TEM' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show() # %% # MCR-1 # ------ target = 'MCR-1' TrainX, TrainY, TestX, TestY, inputs = return_train_test(target, scenario) # %% comparisons = MLClassificationExperiments( input_features=inputs, output_features=target, monitor=['f1_score', 'accuracy', 'precision', 'recall', f1_micro, f1_weighted, precision_micro, precision_weighted, recall_micro, recall_weighted, ], verbosity=0, show=False ) # %% comparisons.fit(x=TrainX.values, y=TrainY.values, exclude=['LinearDiscriminantAnalysis', 'BaggingClassifier', 'HistGradientBoostingClassifier', 'DecisionTreeClassifier', 'GaussianProcessClassifier'] ) # %% _ = comparisons.compare_errors( 'f1_score', x=TestX, y=TestY.values, colors=['salmon', 'cadetblue'], label_bars=True, bar_label_kws={"color": 'black', 'label_type':'center'}, figsize=(10, 8)) plt.tight_layout() if SAVE: plt.savefig(f"results/figures/exp_{target}_{scenario}", bbox_inches="tight", dpi=600) plt.show()