Update app.py

app.py

@@ -1,21 +1,27 @@
 import streamlit as st
 import pandas as pd
 import plotly.express as px
+from transformers import pipeline
 
 # Set the page layout for Streamlit
 st.set_page_config(layout="wide")
 
+# Initialize TAPAS pipeline for table-based question answering
+tqa = pipeline(task="table-question-answering", 
+              model="google/tapas-large-finetuned-wtq",
+              device=0)  # Assuming GPU is available, otherwise set device="cpu"
+
 # Title and Introduction
-st.title("Data Visualization App")
+st.title("Data Visualization App with TAPAS NLP Integration")
 st.markdown(""" 
-    This app allows you to upload a table (CSV or Excel) and ask for graphs to visualize the data.
-    Based on your question, the app will generate interactive graphs using **Plotly**.
+    This app allows you to upload a table (CSV or Excel) and ask questions to generate graphs visualizing the data.
+    Using **TAPAS**, the app can interpret your questions and generate the corresponding graphs.
 
     ### Available Features:
-    - **Graph of a column**: Visualize a column as a graph.
-    - **Scatter plot**: Visualize relationships between two columns.
+    - **Scatter Plot**: Visualize relationships between two columns.
+    - **Line Graph**: Visualize a single column over time.
     
-    Upload your data and ask questions to generate visualizations.
+    Upload your data and ask questions about the data to generate visualizations.
 """)
 
 # File uploader in the sidebar
@@ -36,31 +42,35 @@ else:
             df = None
 
         if df is not None:
-            numeric_columns = df.select_dtypes(include=['object']).columns
-            for col in numeric_columns:
-                df[col] = pd.to_numeric(df[col], errors='ignore')
+            # Convert object columns to numeric where possible
+            df = df.apply(pd.to_numeric, errors='ignore')
 
             st.write("Original Data:")
             st.write(df)
 
-            df_numeric = df.copy()
-            df = df.astype(str)
-
-            # Display the first 5 rows of the dataframe in an editable grid
+            # Display a sample of data for graph generation
             st.write("Sample data for graph generation:")
             st.write(df.head())
 
     except Exception as e:
         st.error(f"Error reading file: {str(e)}")
 
-    # User input for the graph query
+    # User input for the question
     question = st.text_input('Ask your graph-related question')
 
     with st.spinner():
         if st.button('Generate Graph'):
             try:
+                # Use TAPAS model to process the question
+                result = tqa(table=df, query=question)
+
+                # Check if TAPAS is returning column names that could be used for graphing
+                answer = result['answer']
+                st.write(f"TAPAS Answer: {answer}")
+                
+                # Determine if the question relates to graph generation
                 if 'between' in question.lower() and 'and' in question.lower():
-                    # Handle scatter plot (graph between two columns)
+                    # This is a request for a scatter plot (two columns)
                     columns = question.split('between')[-1].split('and')
                     columns = [col.strip() for col in columns]
                     if len(columns) == 2 and all(col in df.columns for col in columns):
@@ -70,7 +80,7 @@ else:
                     else:
                         st.warning("Columns not found in the dataset.")
                 elif 'column' in question.lower():
-                    # Handle graph of a single column
+                    # This is a request for a line graph (single column)
                     column = question.split('of')[-1].strip()
                     if column in df.columns:
                         fig = px.line(df, x=df.index, y=column, title=f"Graph of column '{column}'")
@@ -82,4 +92,4 @@ else:
                     st.warning("Please ask a valid graph-related question (e.g., 'make a graph between column1 and column2').")
 
             except Exception as e:
-                st.warning(f"Error processing question or generating graph: {str(e)}")
+                st.warning(f"Error processing question or generating graph: {str(e)}")