Update app.py

app.py

@@ -21,6 +21,54 @@ class StreamLitResponse(ResponseParser):
 
 gemini_api_key = os.environ['Gemini']
 
+def calculate_kpis(df):
+    """
+    Calculates key performance indicators from a given transaction dataset.
+
+    Args:
+        df: Pandas DataFrame containing transaction data.
+
+    Returns:
+        A JSON object containing the calculated KPIs.
+    """
+
+    # Calculate Total Revenue
+    total_revenue = df['Price'] * df['Quantity'].sum()
+
+    # Calculate Top Five Products by Revenue
+    if df['Description'].nunique() > 5:
+        top_five_products = df.groupby('Description')['Price'].sum().nlargest(5).index.tolist()
+    else:
+        top_five_product = "there are less than 5 products in this dataset"
+
+    if df['Branch'].nunique() > 1:
+        best_branch = df.groupby('Name')['Price'].sum().nlargest(1).index.tolist()
+    else:
+        best_branch = "there is only one branch in this dataset"
+        
+    # Calculate Average Order Value (AOV)
+    aov = df.groupby('Receipt No_')['Price'].sum().mean()
+
+    # Calculate Customer Purchase Frequency (Requires more data for accurate calculation)
+    # Assuming 'Member Card No_' is a unique identifier for customers
+    customer_purchase_frequency = df.groupby('Name.1')['Receipt No_'].nunique().mean()
+
+    # Calculate Estimated Customer Lifetime Value (CLTV) (Requires more data for accurate calculation)
+    # Assuming a simple CLTV model based on AOV and purchase frequency
+    estimated_cltv = aov * customer_purchase_frequency * 12  # Assuming annual value
+
+    # Create JSON output
+    kpis = {
+        "total_revenue": total_revenue,
+        "top_five_products": top_five_products,
+        "average_order_value": aov,
+        "customer_purchase_frequency": customer_purchase_frequency,
+        "estimated_cltv": estimated_cltv,
+        "best_performing_branch": best_branch
+    }
+
+    return json.dumps(kpis, indent=4)
+
 def generateResponse(dataFrame,prompt):
         llm = GoogleGemini(api_key=gemini_api_key)
         pandas_agent = SmartDataframe(dataFrame,config={"llm":llm, "response_parser":StreamLitResponse})