UnboundLocalError：分配前引用的局部變量'graph_df'

Question

該代碼獲取係數並使用它來修改以下DF。它使用另一個函數scaling（）並且工作正常。一旦這個代碼運行，如果我得到它打印的最終結果，new_df它是成功，但未能返回我new_df並引發錯誤，如下所示：

def scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date): 

scaling_dict = scaling(normalised_graph_points_on_xaxis(company_name, company_code, start_date, end_date), 
         coefficient, delta, company_name) 

print(scaling_dict) 

sub_mse = scaling_dict.get('Sub_MSE') 
add_mse = scaling_dict.get('Add_MSE') 

if delta > 0.0001: 

    if abs(add_mse-sub_mse) > 0.05: 

     if add_mse > sub_mse: 
      coefficient = coefficient + delta 
      scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 

     elif add_mse < sub_mse: 
      coefficient = coefficient - delta 
      scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 

    else: 
     delta = delta/2 
     scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 
else: 

    new_coefficient = coefficient 

    print('Co-efficient:', new_coefficient) 

    graph_df = normalised_graph_points_on_xaxis(company_name, company_code, start_date, end_date) 

    print(graph_df) 

    for index, row in graph_df.iterrows(): 
     new_df = graph_df 
     graph_df.set_value(index, twitter_sentiment, row[twitter_sentiment] * coefficient) 

return new_df 

錯誤

File "/Users/Pankaj/PycharmProjects/untitled/scaling_sentiment_graph.py", line 52, in scale_for_all_companies 
scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 
File "/Users/Pankaj/PycharmProjects/untitled/scaling_sentiment_graph.py", line 67, in scale_for_all_companies 
return graph_df 
UnboundLocalError: local variable 'graph_df' referenced before assignment 

我看過類似這樣的帖子，但沒有一個迎合這個特殊情況。請幫忙！謝謝。

Answer 1

試試吧。

def scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date): 

    scaling_dict = scaling(normalised_graph_points_on_xaxis(company_name, company_code, start_date, end_date), 
          coefficient, delta, company_name) 

    print(scaling_dict) 

    sub_mse = scaling_dict.get('Sub_MSE') 
    add_mse = scaling_dict.get('Add_MSE') 

    if delta > 0.0001: 

     if abs(add_mse-sub_mse) > 0.05: 

      if add_mse > sub_mse: 
       coefficient = coefficient + delta 
       return scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 

      elif add_mse < sub_mse: 
       coefficient = coefficient - delta 
       return scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 

     else: 
      delta = delta/2 
      return scale_for_all_companies(coefficient, delta, company_name, company_code, start_date, end_date) 
    else: 

     new_coefficient = coefficient 

     print('Co-efficient:', new_coefficient) 

     graph_df = normalised_graph_points_on_xaxis(company_name, company_code, start_date, end_date) 

     print(graph_df) 

     for index, row in graph_df.iterrows(): 
      new_df = graph_df 
      graph_df.set_value(index, twitter_sentiment, row[twitter_sentiment] * coefficient) 

    return new_df 

而且最好是做兩兩件事：其一，不知何故new_df初始化graph_df.iterrows()之前，以確保如果graph_df.iterrows()是空的，你就不會再崩潰。其次，檢查if abs(add_mse-sub_mse) > 0.05是否可以是False，所以遞歸不會永遠運行。

一件事是，

new_df = graph_df 
graph_df.set_value(index, twitter_sentiment, row[twitter_sentiment] * coefficient)` 

您new_df後改變graph_df，然後返回new_df，所以你要graph_df變化是沒有用的。檢查這一個。也許你想先把graph_df.set_value然後new_df = graph_df？

希望，這將有所幫助。