Update app.py

app.py

@@ -1,13 +1,11 @@
 import gradio as gr
 import pandas as pd
-from transformers import pipeline  # Using Hugging Face LLM
 import matplotlib.pyplot as plt
 import seaborn as sns
 import io
+from transformers import pipeline
 
-# Load Hugging Face's text generation model
-generator = pipeline("text-generation", model="mistralai/Mistral-7B-v0.1")
-
+# Use an open-source model that doesn't require authentication
 def analyze_dataset(file):
     # Load dataset
     df = pd.read_csv(file)
@@ -17,25 +15,52 @@ def analyze_dataset(file):
     missing_values = df.isnull().sum().to_string()
     duplicates = df.duplicated().sum()
     
-    # Prompt LLM for insights
-    prompt = f"""
-    Given the following dataset summary:
-    {summary}
-    Missing Values:
-    {missing_values}
-    Duplicate Entries: {duplicates}
+    # Prepare analysis text
+    insights = f"""Dataset Analysis:
     
-    Provide a structured analysis, visualization suggestions, and cleaning strategies.
-    """
-    response = generator(prompt, max_length=500)
-    insights = response[0]['generated_text']
+Summary Statistics:
+{summary}
+
+Missing Values:
+{missing_values}
+
+Duplicate Entries: {duplicates}
+
+Recommended Cleaning Strategies:
+1. Handle missing values through imputation or removal
+2. Remove or investigate duplicate entries
+3. Consider normalizing numerical features
+4. Check for outliers in the dataset
+"""
     
     # Generate visualizations
-    fig, ax = plt.subplots(figsize=(6, 4))
-    sns.heatmap(df.corr(), annot=True, cmap='coolwarm', ax=ax)
+    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))
+    
+    # Correlation heatmap
+    try:
+        sns.heatmap(df.select_dtypes(include=['float64', 'int64']).corr(), 
+                    annot=True, cmap='coolwarm', ax=ax1)
+        ax1.set_title('Correlation Heatmap')
+    except Exception as e:
+        ax1.text(0.5, 0.5, f"Correlation plot error: {str(e)}", 
+                 horizontalalignment='center', verticalalignment='center')
+    
+    # Distribution plot for numerical columns
+    numerical_cols = df.select_dtypes(include=['float64', 'int64']).columns
+    if len(numerical_cols) > 0:
+        df[numerical_cols].hist(ax=ax2, bins=15)
+        ax2.set_title('Numerical Features Distribution')
+    else:
+        ax2.text(0.5, 0.5, "No numerical columns for distribution", 
+                 horizontalalignment='center', verticalalignment='center')
+    
+    plt.tight_layout()
+    
+    # Save plot to buffer
     buf = io.BytesIO()
     plt.savefig(buf, format='png')
     buf.seek(0)
+    plt.close()
     
     return insights, buf
 
@@ -43,9 +68,13 @@ def analyze_dataset(file):
 demo = gr.Interface(
     fn=analyze_dataset,
     inputs=gr.File(type="filepath"),
-    outputs=[gr.Textbox(label="Analysis"), gr.Image(label="Correlation Heatmap")],
-    title="LLM-Powered Data Analyzer",
-    description="Upload a dataset and get automatic insights, visualizations, and cleaning suggestions."
+    outputs=[
+        gr.Textbox(label="Analysis"), 
+        gr.Image(label="Data Visualizations")
+    ],
+    title="Data Analyzer",
+    description="Upload a CSV file for automatic data analysis and visualization."
 )
 
-demo.launch()
+# Launch the interface
+demo.launch(share=True)