Update app.py

app.py

@@ -11,7 +11,7 @@ import time
 from sklearn.ensemble import RandomForestRegressor
 from sklearn.gaussian_process import GaussianProcessRegressor
 from sklearn.gaussian_process.kernels import RBF, ConstantKernel as C
-# 關鍵修改 1: 導入 XGBoost
+# 關鍵修正: 將 XGBoostRegressor 改為 XGBRegressor
 from xgboost import XGBRegressor
 
 # --- 1. 定義資料與模型設定 ---
@@ -20,7 +20,6 @@ FEATURES = ['MW01', 'MW02', 'MW03', 'flux', 'MWA']
 TARGET = 'yield'
 
 # 檔案名稱用於保存/載入模型和 SHAP 數據
-# 關鍵修改 2: 增加 XGBoost 配置
 MODEL_CONFIG = {
     'Random Forest (RF)': {
         'model_file': "rf_model.joblib",
@@ -94,14 +93,14 @@ def train_and_save_models():
     for name, config in MODEL_CONFIG.items():
         print(f"\n--- Processing Model: {name} ---")
         
-        # 關鍵修改 3: 增加 XGBoost 訓練邏輯
         if name == 'Random Forest (RF)':
             model = RandomForestRegressor(n_estimators=100, random_state=42)
         elif name == 'Gaussian Process (GP)':
             kernel = C(1.0, (1e-3, 1e3)) * RBF(10, (1e-2, 1e2))
             model = GaussianProcessRegressor(kernel=kernel, random_state=42, normalize_y=True)
+        # 關鍵修正: 使用 XGBRegressor
         elif name == 'XGBoost (XGB)':
-            model = XGBoostRegressor(n_estimators=100, random_state=42, use_label_encoder=False, eval_metric='rmse')
+            model = XGBRegressor(n_estimators=100, random_state=42, use_label_encoder=False, eval_metric='rmse')
         else:
             continue
             
@@ -116,7 +115,6 @@ def train_and_save_models():
         print(f"Calculating SHAP values for {name}...")
         try:
             if config['explainer_type'] == shap.TreeExplainer:
-                # TreeExplainer 適用於 RF 和 XGBoost
                 explainer = shap.TreeExplainer(model)
                 shap_values = explainer.shap_values(X)
                 expected_value = explainer.expected_value
@@ -157,7 +155,6 @@ def get_model_explainer_and_data(model_name):
         shap_values = shap_data['shap_values']
         base_value = shap_data['base_value']
         
-        # 即時創建 Explainer
         if config['explainer_type'] == shap.TreeExplainer:
             explainer = shap.TreeExplainer(model)
         elif config['explainer_type'] == shap.KernelExplainer:
@@ -370,10 +367,8 @@ def get_global_decision_plot(model_name):
     X_sample = X.loc[sample_indices]
     
     try:
-        # 使用位置索引
         shap_values_sample = shap_values_full[sample_indices]
     except:
-        # 如果索引不匹配，重新計算（較慢但安全）
         if isinstance(explainer, shap.KernelExplainer):
             shap_values_sample = explainer.shap_values(X_sample.values)
         else:
@@ -413,7 +408,7 @@ input_components = [
     gr.Slider(label="MWA", minimum=5, maximum=20, step=5, value=5)
 ]
 
-# A. 局部解釋 (多圖) - 使用 Blocks (不變)
+# A. 局部解釋 (多圖) - 使用 Blocks
 with gr.Blocks() as local_block:
     gr.Markdown("## Local Explanation for Single Prediction")
     
@@ -479,7 +474,7 @@ with gr.Blocks() as local_block:
     )
 
 
-# B. 全域解釋 (Summary + Decision Plot) - 使用 Blocks (不變)
+# B. 全域解釋 (Summary + Decision Plot) - 使用 Blocks
 with gr.Blocks() as global_summary_block:
     gr.Markdown("## Global Analysis: Feature Importance & Decision Paths")
     
@@ -517,7 +512,7 @@ with gr.Blocks() as global_summary_block:
     )
 
 
-# C. 依賴圖介面 - 使用 Interface (不變)
+# C. 依賴圖介面 - 使用 Interface
 dependence_interface = gr.Interface(
     fn=get_dependence_plot,
     inputs=[