-

机器学习的作业 import pandas as pd from imblearn.over_sampling import SMOTE from sklearn.ensemble import RandomForestClassifier from sklearn.metrics import f1_score, accuracy_score from sklearn.model_selection import train_test_split, RandomizedSearchCV from sklearn.preprocessing import StandardScaler from scipy.stats import randint

导入糖尿病数据集

data = pd.read_csv("E:\excel\data.csv")

对以下字段四舍五入

columns_to_round = ['HighBP', 'HighChol', 'Smoker', 'Stroke', 'Fruits', 'HeartDiseaseorAttack', 'PhysActivity', 'Veggies', 'HvyAlcoholConsump', 'AnyHealthcare', 'NoDocbcCost', 'DiffWalk', 'Sex', 'Education'] data[columns_to_round] = data[columns_to_round].round().astype(int)

数据预处理

data_X = data.drop(['target', 'id'], axis=1) # 去除target,id两列 data_y = data['target']

拆分测试集和训练集

X_train, X_test, y_train, y_test = train_test_split(data_X, data_y, test_size=0.2, random_state=42)

SMOTE: 合成少数类过采样技术

smote = SMOTE(random_state=42) X_train_resampled, y_train_resampled = smote.fit_resample(X_train, y_train)

数据标准化

scaler = StandardScaler() X_train_resampled = scaler.fit_transform(X_train_resampled) X_test = scaler.transform(X_test)

随机森林模型

rf = RandomForestClassifier(random_state=42)

定义要调优的超参数范围

param_dist = { 'n_estimators': randint(50, 200), # 树的数量 'max_depth': randint(5, 20), # 最大深度 'min_samples_split': randint(2, 10), # 最小分裂样本数 'min_samples_leaf': randint(1, 5), # 最小叶子节点样本数 'max_features': [None, 'sqrt', 'log2'], 'class_weight': ['balanced', None], }

使用RandomizedSearchCV进行参数调优，选择最佳参数

grid_search = RandomizedSearchCV(rf, param_distributions=param_dist, n_iter=50, cv=5, scoring='f1_macro', n_jobs=-1, random_state=42) grid_search.fit(X_train_resampled, y_train_resampled)

输出最佳参数和对应的F1分数

print(f"Best parameters: {grid_search.best_params_}") print(f"Best F1 score: {grid_search.best_score_:.3f}")

获取最佳模型

best_rf = grid_search.best_estimator_

进行预测

y_pred = best_rf.predict(X_test)

计算 F1 分数和准确度

f1 = f1_score(y_test, y_pred, average='macro') accuracy = accuracy_score(y_test, y_pred)

输出结果

print(f"F1 score: {f1:.3f}") print(f"Accuracy score: {accuracy:.3f}")