'C': 0.001,
'verbose': 0
})
glm_anova.fit(X_train_anova_p, y_train_p)
glm_anova_raw_preds, glm_anova_preds, glm_anova_score = glm_anova.predict(
X_test_anova_p, y_test_p)
glm_anova_fig = Classifiers.metrics(y_test_p, glm_anova_raw_preds[:, 1],
glm_anova_preds)
################################
######## XGB ROC Curves ########
xg_total = Classifiers.XGBoostModel(params={
'booster': 'gbtree',
'eta': 0.3,
'max_depth': 6
})
xg_total.fit(X_train_p, y_train_p)
xg_total_raw_preds, xg_total_preds, xg_total_score = xg_total.predict(
X_test_p, y_test_p)
xg_total_fig = Classifiers.metrics(y_test_p, xg_total_raw_preds[:, 1],
xg_total_preds)
xg_vt = Classifiers.XGBoostModel(params={
'booster': 'gbtree',
'eta': 0.3,
'max_depth': 6
})
# 'C' : 0.001,
"verbose": 0,
}
)
glm_anova.fit(X_train_p, y_train_p)
glm_anova_raw_preds, glm_anova_preds, glm_anova_score = glm_anova.predict(
X_test_p, y_test_p
)
auc_glm, pr_auc_glm, fpr_glm, tpr_glm, roc_thresholds_glm, recalls_glm, precisions_glm = get_auc_pr(
y_test_p, glm_anova_raw_preds
)
xgb = Classifiers.XGBoostModel(params={**best_params_xg})
xgb.fit(X_train_p, y_train_p)
xgb_raw_preds, xgb_preds, xgb_score = xgb.predict(X_test_p, y_test_p)
auc_xgb, pr_auc_xgb, fpr_xgb, tpr_xgb, roc_thresholds_xgb, recalls_xgb, precisions_xgb = get_auc_pr(
y_test_p, xgb_raw_preds
)
recall_no_skill = y_train_p.sum() / len(y_train_p)
optimal_preds, threshold = optimal_points(fpr, tpr, rf_anova_raw_preds, roc_thresholds)
optimal_preds_glm, threshold_glm = optimal_points(
fpr_glm, tpr_glm, glm_anova_raw_preds, roc_thresholds_glm
# Save model to disk
filename = 'S:/Dehydration_stroke/Team Emerald/Working GitHub Directories/'\
'Michael/stroke-hemodynamics/Aim 2/Models/FullModelResults/UpdatedResults/TEST24hr_model_rf.sav'
with open(filename, 'rb') as f:
rf = pickle.load(f)
rf_raw_preds, rf_preds, rf_score = rf.predict(X_test_p, y_test_p)
auc_rf, pr_auc_rf, fpr, tpr, roc_thresholds, recalls, precisions = get_auc_pr(
y_test_p, rf_raw_preds
)
xgb = Classifiers.XGBoostModel(
params={"booster": "gblinear", "eta": 0.1, "max_depth": 8}
)
xgb.fit(X_train_p, y_train_p)
# Save model to disk
filename = 'S:/Dehydration_stroke/Team Emerald/Working GitHub Directories/'\
'Michael/stroke-hemodynamics/Aim 2/Models/FullModelResults/UpdatedResults/TEST24hr_model_xgb.sav'
with open(filename, 'rb') as f:
xgb = pickle.load(f)
xgb_raw_preds, xgb_preds, xgb_score = xgb.predict(X_test_p, y_test_p)
auc_xgb, pr_auc_xgb, fpr_xgb, tpr_xgb, roc_thresholds_xgb, recalls_xgb, precisions_xgb = get_auc_pr(
y_test_p, xgb_raw_preds
)
