Update app.py

app.py

@@ -539,6 +539,61 @@ def upload_and_select_dataframe():
       st.sidebar.info("Please upload some files.")
       return None
 
+def categorize_responses(df, api_key, prompt=None):
+    """
+    Categorize responses in a DataFrame using OpenAI's GPT-based API.
+
+    Args:
+        df (pd.DataFrame): Input DataFrame with columns 'id' and 'response'.
+        api_key (str): OpenAI API key for accessing GPT-based models.
+        prompt (str, optional): A question or explanation of the survey question for better categorization.
+
+    Returns:
+        pd.DataFrame: DataFrame with 'id' and 'category' columns.
+    """
+    # Set OpenAI API key
+    openai.api_key = api_key
+
+    # Combine the question prompt with instructions for categorization
+    base_prompt = "Categorize the following responses into conceptual categories."
+    if prompt:
+        base_prompt = f"{prompt}\n\n{base_prompt}"
+
+    # Initialize a list to store results
+    categories = []
+
+    # Iterate through the responses in the DataFrame
+    for response in df['response']:
+        # Construct the prompt for the specific response
+        full_prompt = f"{base_prompt}\n\nResponse: {response}\n\nCategory:"
+
+        try:
+            # Use the OpenAI API to get the category
+            completion = openai.Completion.create(
+                engine="text-davinci-003",  # Use a GPT-3.5 engine or similar
+                prompt=full_prompt,
+                max_tokens=50,  # Limit the token count for category output
+                temperature=0.7
+            )
+
+            # Extract the category from the API response
+            category = completion.choices[0].text.strip()
+        except Exception as e:
+            # Handle exceptions (e.g., API issues) and assign a default value
+            print(f"Error processing response: {response}\n{e}")
+            category = "Unknown"
+
+        # Append the category to the results list
+        categories.append(category)
+
+    # Create a new DataFrame with IDs and categories
+    categorized_df = pd.DataFrame({
+        'id': df['id'],
+        'category': categories
+    })
+
+    return categorized_df
+
 empty_col1, main_col, empty_col2 = st.columns([1.6, 2.8, 1.6])
 
 with main_col:
@@ -550,8 +605,6 @@ with main_col:
     with col2:
         st.title("Chortke")
 
-import streamlit as st
-
 st.markdown('[Click to register a suggestion or comment](https://docs.google.com/forms/d/e/1FAIpQLScLyP7bBbqMfGdspjL7Ij64UZ6v2KjqjKNbm8gwEsgWsFs_Qg/viewform?usp=header)')
 
 df = upload_and_select_dataframe()
@@ -888,8 +941,21 @@ try:
                 result = analyze_z_test(uploaded_file)
                 if result:
                     st.success("Z-Test analysis completed successfully.")
-    
-    elif main_option in ["Machine Learning", "Coding"]:
+
+    elif main_option == "Coding":
+        selected_list = st.sidebar.multiselect(
+            'Select the desired "Open Question" column.', 
+            cols,
+            default=[]
+        )
+        api_key = "sk-1a0127fbc52a4e50a93bd5cd18af3a85"
+        prompt_user = st.text_input("Write a brief description of the selected column question.")
+        if st.button("Submit"):
+            df2 = categorize_responses(df[selected_list], api_key, prompt=prompt_user)
+            st.subheader("Categorized data")
+            st.dataframe(df2)
+        
+    elif main_option == "Machine Learning":
         st.info("This section of the program is under development.")
 
 except Exception as e: