### 编写 Python 实现随机森林算法筛选最优盐分指数 以下是一个完整的解决方案，展示如何使用随机森林算法根据土壤电导率、pH 值和不同类型的盐分指数数据，筛选出最优的三种盐分指数。 --- #### 数据准备与加载 假设已有一个包含土壤电导率 (`conductivity`)、pH 值 (`ph`) 和若干种盐分指数的数据集。这些数据存储在一个 CSV 文件中，其中每一列代表一种特征或目标变量。 ```python import pandas as pd from sklearn.ensemble import RandomForestClassifier from sklearn.model_selection import train_test_split from sklearn.metrics import accuracy_score, classification_report # 加载数据 df = pd.read_csv('soil_data.csv') # 查看数据结构 print(df.head()) # 定义特征和目标变量 features = ['conductivity', 'ph'] targets = [col for col in df.columns if 'salt_index_' in col] # 构造新的 DataFrame 来表示每个样本属于哪种盐分指数类别 def create_target_column(row): max_value = row.max() return targets[row.tolist().index(max_value)] target_df = df[targets] df['best_salt_index'] = target_df.apply(create_target_column, axis=1) X = df[features] y = df['best_salt_index'] # 划分训练集和测试集 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) ``` --- #### 随机森林模型构建与训练 使用 `RandomForestClassifier` 对数据进行建模，并通过交叉验证优化超参数。 ```python # 初始化随机森林分类器 rf_model = RandomForestClassifier(n_estimators=100, random_state=42) # 训练模型 rf_model.fit(X_train, y_train) # 测试模型性能 y_pred = rf_model.predict(X_test) accuracy = accuracy_score(y_test, y_pred) report = classification_report(y_test, y_pred) print(f"Model Accuracy: {accuracy:.4f}") print(report) ``` --- #### 特征重要性分析 通过分析随机森林模型中的特征重要性分数，识别哪些输入变量对预测结果的影响最大。 ```python # 输出特征的重要性 feature_importances = rf_model.feature_importances_ for feature_name, importance in zip(features, feature_importances): print(f"{feature_name}: {importance:.4f}") # 可视化特征重要性 import matplotlib.pyplot as plt plt.barh(features, feature_importances, color='skyblue') plt.xlabel("Feature Importance Score") plt.ylabel("Features") plt.title("Feature Importances from Random Forest Model") plt.show() ``` --- #### 筛选出最优的三种盐分指数 通过对模型预测的概率分布进行排序，提取概率最高的三个盐分指数作为最优解。 ```python # 获取测试集中每个样本的最佳盐分指数及其概率 probabilities = rf_model.predict_proba(X_test) top_indices = probabilities.argsort(axis=1)[:, -3:] # 每个样本按概率从大到小取前三名 # 将索引映射回原始盐分指数名称 inverse_mapping = dict(zip(range(len(targets)), targets)) top_three_indices = [[inverse_mapping[i] for i in indices][::-1] for indices in top_indices] # 打印前几个样本的结果 for i, sample_top_three in enumerate(top_three_indices[:5]): print(f"Sample {i+1} Top Three Salt Indices: {sample_top_three}") ``` --- #### 总结代码片段 将以上各部分组合起来形成一个完整的脚本文件，便于执行和调试。 ```python if __name__ == "__main__": # 数据加载与预处理 df = pd.read_csv('soil_data.csv') features = ['conductivity', 'ph'] targets = [col for col in df.columns if 'salt_index_' in col] def create_target_column(row): max_value = row.max() return targets[row.tolist().index(max_value)] target_df = df[targets] df['best_salt_index'] = target_df.apply(create_target_column, axis=1) X = df[features] y = df['best_salt_index'] X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) # 模型训练与评估 rf_model = RandomForestClassifier(n_estimators=100, random_state=42) rf_model.fit(X_train, y_train) y_pred = rf_model.predict(X_test) accuracy = accuracy_score(y_test, y_pred) report = classification_report(y_test, y_pred) print(f"Model Accuracy: {accuracy:.4f}") print(report) # 特征重要性分析 feature_importances = rf_model.feature_importances_ for feature_name, importance in zip(features, feature_importances): print(f"{feature_name}: {importance:.4f}") # 筛选最优盐分指数 probabilities = rf_model.predict_proba(X_test) top_indices = probabilities.argsort(axis=1)[:, -3:] inverse_mapping = dict(zip(range(len(targets)), targets)) top_three_indices = [[inverse_mapping[i] for i in indices][::-1] for indices in top_indices] for i, sample_top_three in enumerate(top_three_indices[:5]): print(f"Sample {i+1} Top Three Salt Indices: {sample_top_three}") ``` --- ###