test2

app.py

@@ -19,25 +19,21 @@ def prepare_data_chronos(data):
     formatted_df = pd.DataFrame({
         'item_id': ['stock'] * len(df),
         'timestamp': pd.to_datetime(df['Date']),
-        'value': df['Close'].astype('float32').values.ravel()  # 確保是 1-dimensional
+        'value': df['Close'].astype('float32').values.ravel()
     })
     
     # 按照時間戳排序
     formatted_df = formatted_df.sort_values('timestamp')
     
     try:
-        # 直接創建 TimeSeriesDataFrame
-        ts_df = TimeSeriesDataFrame(formatted_df)
-        
-        # 打印一些診斷信息
-        print("TimeSeriesDataFrame shape:", ts_df.shape)
-        print("Value column shape:", ts_df['value'].shape)
-        
+        # 指定頻率創建 TimeSeriesDataFrame
+        ts_df = TimeSeriesDataFrame(
+            formatted_df,
+            freq="1D"  # 指定為每日數據
+        )
         return ts_df
     except Exception as e:
         print(f"Error creating TimeSeriesDataFrame: {str(e)}")
-        print("Formatted DataFrame info:")
-        print(formatted_df.info())
         raise
 
 
@@ -113,21 +109,19 @@ def get_top_10_potential_stocks(period, selected_indices):
     
     for ticker in stock_list:
         try:
-            # 獲取股票數據
             data = get_stock_data(ticker, period)
             if data.empty:
                 continue
                 
-            # 準備數據
             ts_data = prepare_data_chronos(data)
             
-            # 創建預測器
+            # 創建預測器時指定頻率
             predictor = TimeSeriesPredictor(
                 prediction_length=prediction_length,
-                target='value'  # 指定目標列名
+                target='value',
+                freq="1D"  # 指定為每日數據
             )
             
-            # 訓練模型
             predictor.fit(
                 ts_data,
                 hyperparameters={
@@ -135,10 +129,7 @@ def get_top_10_potential_stocks(period, selected_indices):
                 }
             )
             
-            # 進行預測
             predictions = predictor.predict(ts_data)
-            
-            # 計算潛力
             potential = (predictions.iloc[-1] - data['Close'].iloc[-1]) / data['Close'].iloc[-1]
             stock_predictions.append((ticker, potential, data['Close'].iloc[-1], predictions.iloc[-1]))