Update app.py

app.py

@@ -1,162 +1,142 @@
 import streamlit as st
 import pandas as pd
+import numpy as np
 import os
 import matplotlib.pyplot as plt
 import seaborn as sns
-from dotenv import load_dotenv
-from io import BytesIO
-
-# AI & ML Components
+from sklearn.cluster import KMeans
+from statsmodels.tsa.holtwinters import ExponentialSmoothing
 from langchain_openai import ChatOpenAI
-from pandasai import SmartDatalake
-from pandasai.llm import OpenAI
-from pycaret.regression import setup as setup_reg, compare_models, predict_model, pull, save_model, load_model
+from dotenv import load_dotenv
 
 # ✅ Load Environment Variables
 load_dotenv()
 openai_api_key = os.getenv("OPENAI_API_KEY")
 
 if not openai_api_key:
-    st.error("❌ OpenAI API Key is missing! Set it in your .env file.")
+    st.error("❌ OpenAI API Key is missing! Set it in your environment variables.")
 
-# ✅ Streamlit Page Configuration
-st.set_page_config(page_title="AI-Powered Store Recommender & Chatbot", layout="wide")
+# ✅ Auto-Load All Files from the Folder
+folder_path = os.getcwd()  # Uses current folder
 
-# ✅ Title
-st.title("🛍️ AI-Powered Apparel Group Store Recommender & Chatbot")
+# Load KPI Excel File (Main store dataset)
+data_file = os.path.join(folder_path, "kpi.xlsx")
+df = pd.read_excel(data_file, engine="openpyxl")
 
-# ✅ Load KPI & Competitor Data from Repo
-repo_path = os.getcwd()
-kpi_file = os.path.join(repo_path, "kpi.xlsx")
+# Load Competitor Mapping Files
 competitor_files = {
-    "DEIRA CITY CENTRE": "Competitor Mapping Doc Final1(DEIRA CITY CENTRE).csv",
-    "ARABIAN CENTRE": "Competitor Mapping Doc Final1(ARABIAN CENTRE).csv",
-    "AL GHURAIR CENTRE": "Competitor Mapping Doc Final1(AL GHURAIR CENTRE).csv",
-    "IBN BATTUTA MALL": "Competitor Mapping Doc Final1(IBN BATTUTA MALL).csv",
-    "DUBAI HILLS": "Competitor Mapping Doc Final1(DUBAI HILLS).csv"
+    "Deira City Centre": os.path.join(folder_path, "Competitor Mapping Doc Final1(DEIRA CITY CENTRE).csv"),
+    "Arabian Centre": os.path.join(folder_path, "Competitor Mapping Doc Final1(ARABIAN CENTRE).csv"),
+    "Al Ghurair Centre": os.path.join(folder_path, "Competitor Mapping Doc Final1(AL GHURAIR CENTRE).csv"),
+    "Ibn Battuta Mall": os.path.join(folder_path, "Competitor Mapping Doc Final1(IBN BATTUTA MALL).csv"),
+    "Dubai Hills": os.path.join(folder_path, "Competitor Mapping Doc Final1(DUBAI HILLS).csv"),
 }
 
-# ✅ Load KPI Data
-df_kpi = pd.read_excel(kpi_file, engine="openpyxl")
+# Store all competitor data in a dictionary
+competitor_data = {}
+for mall, file in competitor_files.items():
+    competitor_data[mall] = pd.read_csv(file, encoding="ISO-8859-1")
 
-# ✅ Load Competitor Data with ISO Encoding
-df_competitors = {}
-for mall, file_name in competitor_files.items():
-    file_path = os.path.join(repo_path, file_name)
-    if os.path.exists(file_path):
-        try:
-            df_competitors[mall] = pd.read_csv(file_path, encoding="ISO-8859-1", on_bad_lines="skip")
-        except Exception as e:
-            st.error(f"❌ Error loading {file_name}: {e}")
+# ✅ Streamlit Page Configuration
+st.set_page_config(page_title="AI Store Replacement, Demand Prediction & Chatbot", layout="wide")
+
+# ✅ Title
+st.title("🔄 AI-Powered Store Replacement, Demand Prediction & Chatbot")
 
-# ✅ Display Previews
-st.subheader("📊 KPI Data Preview (5 Unique Rows)")
-st.dataframe(df_kpi.drop_duplicates().sample(n=5, random_state=42) if len(df_kpi) > 5 else df_kpi)
+# ✅ Display Data Preview (5 Random Samples)
+st.subheader("📊 Sample Data Preview from KPI Sheet")
+st.dataframe(df.sample(5))  # Show 5 random unique rows
 
-st.subheader("🏆 Competitor Data Preview (First 5 Rows of Each Mall)")
-for mall, df in df_competitors.items():
-    st.write(f"### {mall} - Competitors")
-    st.dataframe(df.head(5))
+# ✅ Store Replacement System
+st.subheader("🏬 AI-Based Store Replacement")
 
-# ✅ Competitor Analysis for Selected Mall
-st.subheader("🔎 Competitor Analysis")
-selected_mall = st.selectbox("🏬 Select a Mall for Competitor Insights:", df_kpi["Mall"].unique())
+# Select Mall
+selected_mall = st.selectbox("🏬 Select a Mall:", df["Mall"].unique())
 
 if selected_mall:
-    if selected_mall in df_competitors:
-        competitors = df_competitors[selected_mall]["Store_Brand"].unique()
-        st.write(f"💡 **Competitors in {selected_mall}:**")
-        st.write(competitors)
-    else:
-        # ✅ Use AI to Predict Competitors for Malls Not in Dataset
+    # Get all stores in the selected mall
+    mall_data = df[df["Mall"] == selected_mall]
+    
+    # Select the store that will be replaced
+    store_to_replace = st.selectbox("🚪 Select the store that is closing:", mall_data["Store_Brand"].unique())
+
+    if store_to_replace:
         llm = ChatOpenAI(model="gpt-4", api_key=openai_api_key)
-        prompt = f"Based on available data, suggest competitors for {selected_mall}."
-        ai_competitors = llm.invoke(prompt)
-        st.warning(f"⚠️ No direct competitor data found. AI has predicted competitors:")
-        st.write(ai_competitors)
-
-# ✅ AI-Generated Store Recommendations
-st.subheader("🔮 AI-Based Store Recommendations")
-store_to_replace = st.selectbox("🚪 Select a Store to Replace:", df_kpi["Store_Brand"].unique())
-
-if store_to_replace:
-    # ✅ Identify Store Category
-    prompt = f"Identify the category of {store_to_replace} (e.g., Fashion, Electronics, Footwear, etc.)."
-    store_category = llm.invoke(prompt)
-    st.write(f"🔍 **Identified Category:** {store_category}")
-
-    # ✅ Find 3 AI-Recommended Replacement Brands NOT Present in This Mall
-    all_brands = set(df_kpi["Store_Brand"].unique())
-    existing_brands = set(df_kpi[df_kpi["Mall"] == selected_mall]["Store_Brand"].unique())
-    available_replacements = list(all_brands - existing_brands)
-
-    if available_replacements:
-        prompt = f"Suggest 3 alternative brands in {store_category} category that are not present in {selected_mall}. Possible replacements: {available_replacements}"
-        ai_recommendations = llm.invoke(prompt)
-        st.success(f"🔄 **AI Recommended Replacements for {store_to_replace}:**")
-        st.write(ai_recommendations)
-    else:
-        st.warning("⚠️ No new brands found for replacement.")
-
-# ✅ Machine Learning-Based Sales Prediction
-st.subheader("📈 Predict Annual Sales for a New Store")
-selected_brand = st.selectbox("🛍️ Select a Brand:", df_kpi["Store_Brand"].unique())
-
-if selected_brand:
-    # **Step 1: Select Features**
-    if "Annual_Sales" in df_kpi.columns:
-        numeric_cols = df_kpi.select_dtypes(include=["number"]).columns.tolist()
-        feature_cols = st.multiselect("📊 Select Features for Prediction:", numeric_cols, default=numeric_cols)
-
-        if feature_cols:
-            # **Step 2: Train Model (Only Once Per File)**
-            model_save_path = "best_sales_model"
-            if not os.path.exists(model_save_path):
-                st.info("🔄 Training the ML Model... (This may take a few seconds)")
-                setup_reg(df_kpi[feature_cols + ["Annual_Sales"]], target="Annual_Sales", silent=True, session_id=123)
-                best_model = compare_models()
-                save_model(best_model, model_save_path)
-                st.success("✅ Model Training Complete! Best model saved.")
-
-            # **Step 3: Load Trained Model**
-            best_model = load_model(model_save_path)
-
-            # **Step 4: Predict Sales for a New Store**
-            ref_store = df_kpi[df_kpi["Store_Brand"] == selected_brand]
-            if not ref_store.empty:
-                new_store_data = ref_store[feature_cols].mean().to_frame().T  # Use mean as input
-                predicted_sales = predict_model(best_model, data=new_store_data)["Label"].iloc[0]
-
-                st.write(f"📊 **Predicted Annual Sales for {selected_brand}:** **${predicted_sales:,.2f}**")
-
-                # **Feature Importance Chart**
-                st.subheader("🔍 Key Factors Affecting Sales")
-                feature_importance = pull()
-                st.bar_chart(feature_importance)
-
-            else:
-                st.warning(f"⚠️ No reference data available to predict sales for {selected_brand}.")
-
-# ✅ AI Chatbot for Business Insights
-st.subheader("💬 AI Chatbot for Business Insights")
-sdl = SmartDatalake([df_kpi, *df_competitors.values()], config={
-    "llm": OpenAI(api_token=openai_api_key),
-    "use_error_correction_framework": True,
-    "enable_cache": False,
-    "enable_restricted_mode": False,
-    "save_charts": True,
-    "save_charts_path": "/home/user/app/exports/charts/temp_chart.png"
-})
-
-user_query = st.text_input("🔍 Ask AI About the Dataset:")
+
+        # ✅ Identify Store Category
+        category_prompt = f"Given that {store_to_replace} is a retail brand, what type of store is it? Categorize it broadly (e.g., Fashion, Electronics, Footwear, Food, Home Decor, etc.)."
+        store_category = llm.invoke(category_prompt).content
+        st.write(f"🔍 **Identified Category:** {store_category}")
+
+        # ✅ Find 5 Competitors in the Same Mall
+        if selected_mall in competitor_data:
+            mall_competitor_data = competitor_data[selected_mall]
+            competitor_list = mall_competitor_data["Competitors in mall"].dropna().unique()
+
+            if len(competitor_list) > 5:
+                competitor_list = np.random.choice(competitor_list, 5, replace=False)
+
+            st.subheader("🔎 Competitors in This Mall")
+            for competitor in competitor_list:
+                st.write(f"- {competitor}")
+
+        # ✅ AI-Based Store Clustering (Finding Best Replacements)
+        store_features = df[["Annual_Sales", "Footfall", "Gross_Margin"]].dropna()
+        kmeans = KMeans(n_clusters=5, random_state=42)  # 5 Clusters
+        df["Cluster"] = kmeans.fit_predict(store_features)
+
+        # ✅ Find a replacement store in the same category (From kpi.xlsx, Not in this mall)
+        all_stores = set(df["Store_Brand"].unique())
+        existing_stores = set(mall_data["Store_Brand"].unique())
+        potential_replacements = list(all_stores - existing_stores)
+
+        if potential_replacements:
+            replacement_prompt = f"Based on the category {store_category}, suggest 3 stores from the list below that are not already in {selected_mall}. The stores should match the mall’s customer base. Store List: {potential_replacements}"
+            replacement_response = llm.invoke(replacement_prompt).content
+            replacement_stores = replacement_response.split(", ")
+
+            st.success(f"🔄 **Recommended Replacement Stores:**")
+            for replacement_store in replacement_stores:
+                store_data = df[df["Store_Brand"] == replacement_store]
+                
+                if not store_data.empty:
+                    sales_history = store_data["Annual_Sales"]
+                    if len(sales_history) > 3:
+                        # ✅ Time-Series Forecasting for Demand Prediction
+                        model = ExponentialSmoothing(sales_history, trend="add", seasonal="add", seasonal_periods=3)
+                        fit = model.fit()
+                        forecast_sales = fit.forecast(1).values[0]  # Predict next year's sales
+
+                        expected_footfall = store_data["Footfall"].mean()
+                        expected_conversion = store_data["Conversion_Rate"].mean()
+                        expected_margin = store_data["Gross_Margin"].mean()
+
+                        st.write(f"**{replacement_store}**")
+                        st.write(f"📊 **Projected Sales:** AED {forecast_sales:,.2f}")
+                        st.write(f"👣 **Projected Footfall:** {expected_footfall:,.0f} visitors")
+                        st.write(f"💰 **Projected Margin:** AED {expected_margin:,.2f}")
+                        st.write(f"🔄 **Projected Conversion Rate:** {expected_conversion:.2f}%")
+
+                        # ✅ Justification
+                        justification_prompt = f"Explain why {replacement_store} is a good replacement for {store_to_replace} in {selected_mall}. Include numeric KPIs: expected sales, footfall, conversion rate, and profit margins."
+                        justification = llm.invoke(justification_prompt).content
+                        st.write(f"🤖 **AI Justification:** {justification}")
+
+# ✅ AI Chatbot for Dynamic Insights
+st.subheader("💬 AI-Powered Chatbot")
+
+# AI Chatbot using LangChain
+user_query = st.text_input("🔍 Ask anything about store performance, competitors, or pricing:")
 if user_query:
-    with st.spinner("🤖 AI is analyzing..."):
+    with st.spinner("🤖 AI is thinking..."):
         try:
-            ai_response = sdl.chat(user_query)
-            st.markdown(f"🤖 **AI Response:** {ai_response}")
+            chat_prompt = f"Answer the following retail business question based on your market knowledge: {user_query}"
+            chat_response = llm.invoke(chat_prompt).content
+            st.markdown(f"🤖 **AI Response:** {chat_response}")
+
         except Exception as e:
             st.error(f"❌ AI Error: {str(e)}")
 
 # ✅ Footer
 st.markdown("---")
-st.markdown("💡 Built with ❤️ using **Streamlit, PandasAI, PyCaret & GPT-4**")
+st.markdown("💡 Built with ❤️ using **LangChain AI for Store Replacement & Demand Prediction**")