Merge remote-tracking branch 'origin/main' into main

ml/main.py

@@ -18,14 +18,14 @@ def __init__(self, input_size, hidden_size, num_layers, output_size):
         super(LSTMModel, self).__init__()
         self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
         self.fc = nn.Linear(hidden_size, output_size)
-    
+
     def forward(self, x):
         out, _ = self.lstm(x)
         out = self.fc(out)
-        
+
         return out
 
-input_size = 13
+input_size = 12
 hidden_size = 64
 num_layers = 2
 output_size = 4
@@ -46,21 +46,26 @@ async def predict(input_data: str = Form(...),type_num: int = Form(...)):
 
 
     # 모델에 대한 타입 지정을 num 에 해줘야함
-    # 모델의 타입은 직렬 정보에 따라서 1 = 2 = 3 = 4 = 5 = 6 = 
-
-
-
-
+    # 모델의 타입은 직렬 정보에 따라서 1 = 2 = 3 = 4 = 5 = 6 =
     model.load_state_dict(torch.load('lstm_model_'+str(type_num)+'.pth'))
     model.eval()
 
     input_data = ast.literal_eval(input_data)
-    filtered_values = [value for value in input_data if value != -1]
-    mean_value = np.mean(filtered_values)
-
-    input_data = [mean_value if value == -1 else value for value in input_data]
-    input_data.append(sum(input_data))
-
+    # 모두 0이면 리스트에 0 추가
+    if all(value == 0 for value in input_data):
+        input_data.append(0)
+    else:
+        # 0 값을 제외한 리스트 생성
+        filtered_values = [value for value in input_data if value != 0]
+
+        # 0이 없을 때만 평균값 계산 및 처리
+        if filtered_values:
+            mean_value = np.mean(filtered_values)
+            input_data = [mean_value if value == 0 else value for value in input_data]
+            input_data.append(sum(input_data))
+        else:
+            input_data.append(sum(input_data))
+
     print(f"model input data is {input_data}")
 
     result = predict_single_input(input_data)

ml/train.py

@@ -35,13 +35,12 @@ def data_preprocess(df):
 df=data_preprocess(df)
 
 
-
-type_1_df=df[df['applicationType']=='경찰직공무원(여)']
-type_2_df=df[df['applicationType']=='경찰직공무원(남)']
+type_1_df=df[df['applicationType']=='경찰직공무원(남)']
+type_2_df=df[df['applicationType']=='경찰직공무원(여)']
 type_3_df=df[df['applicationType']=='소방직공무원(남)']
-type_4_df=df[df['applicationType']=='교정직공무원(남)']
-type_5_df=df[df['applicationType']=='소방직공무원(여)']
-type_6_df=df[df['applicationType']=='교정직공무원(여)']
+type_4_df=df[df['applicationType']=='소방직공무원(여)']
+type_5_df=df[df['applicationType']=='경호직공무원(남)']
+type_6_df=df[df['applicationType']=='경호직공무원(여)']
 
 type_1_df=type_1_df.drop('applicationType',axis=1)
 type_2_df=type_2_df.drop('applicationType',axis=1)
@@ -50,7 +49,7 @@ def data_preprocess(df):
 type_5_df=type_5_df.drop('applicationType',axis=1)
 type_6_df=type_6_df.drop('applicationType',axis=1)
 
-
+print(type_3_df)
 def convert(df):
 
     data = {
@@ -62,15 +61,15 @@ def convert(df):
 
     # 피벗 테이블 생성
     pivot_table = df.groupby(['member_id', 'month'])['score'].sum().unstack()
-    
+
     return pivot_table
 
-type_1_df=convert(type_1_df)   
-type_2_df=convert(type_2_df)   
-type_3_df=convert(type_3_df)   
-type_4_df=convert(type_4_df)   
-type_5_df=convert(type_5_df)   
-type_6_df=convert(type_6_df)   
+type_1_df=convert(type_1_df)
+type_2_df=convert(type_2_df)
+type_3_df=convert(type_3_df)
+type_4_df=convert(type_4_df)
+type_5_df=convert(type_5_df)
+type_6_df=convert(type_6_df)
 
 
 type_1_df=type_1_df.fillna(type_1_df.mean())
@@ -106,24 +105,32 @@ def set_label(df):
 type_6_df = set_label(type_6_df)
 
 
+type_1_df=type_1_df.drop("total_score",axis=1)
+type_2_df=type_2_df.drop("total_score",axis=1)
+type_3_df=type_3_df.drop("total_score",axis=1)
+type_4_df=type_4_df.drop("total_score",axis=1)
+type_5_df=type_5_df.drop("total_score",axis=1)
+type_6_df=type_6_df.drop("total_score",axis=1)
+
+
 
 
 class LSTMModel(nn.Module):
     def __init__(self, input_size, hidden_size, num_layers, output_size):
         super(LSTMModel, self).__init__()
         self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
         self.fc = nn.Linear(hidden_size, output_size)
-    
+
     def forward(self, x):
-        out, _ = self.lstm(x)
+        out,_= self.lstm(x)
         out = self.fc(out)  # Remove this line
-        
+
         return out
-        
+
 df_list=[type_1_df,type_2_df,type_3_df,type_4_df,type_5_df,type_6_df]
 num=1
 for i in df_list:
-    
+
     # Convert the DataFrame to PyTorch tensors
     data_tensor = torch.tensor(i.drop(columns=['label']).values, dtype=torch.float32)
     label_tensor = torch.tensor(i['label'].values, dtype=torch.long)  # Change dtype to long
@@ -138,7 +145,7 @@ def forward(self, x):
     test_labels = label_tensor[train_size:]
 
     # Create an instance of the LSTMModel
-    input_size = 13
+    input_size = 12
     hidden_size = 64
     num_layers = 2
     output_size = 4
@@ -153,13 +160,13 @@ def forward(self, x):
     for epoch in range(num_epochs):
         model.train()
         optimizer.zero_grad()
-        
+
         outputs = model(train_data)
         loss = criterion(outputs.squeeze(), train_labels)  # Remove the dimension from outputs
-        
+
         loss.backward()
         optimizer.step()
-        
+
         if (epoch + 1) % 10 == 0:
             print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {loss.item():.4f}')