app.py

import streamlit as st
import pandas as pd
import numpy as np
import joblib
import os
import matplotlib.pyplot as plt
import seaborn as sns

st.set_page_config(page_title="Enterprise Customer Insights Hub", layout="wide", page_icon="📊")

# Load model, scaler, and source data with performance caching
@st.cache_resource
def load_production_assets():
    model = joblib.load('models/kmeans_model.pkl')
    scaler = joblib.load('models/scaler.pkl')
    raw_df= pd.read_csv('data/Mall_Customers.csv')
    return model, scaler, raw_df

# Fail-safe check for missing trained model files
try:
    model, scaler, raw_df = load_production_assets()
    X_scaled = scaler.transform(raw_df.iloc[:, [3, 4]].values)
    raw_df['Cluster'] = model.predict(X_scaled)
except Exception as e:
    st.error(" System Error: Please execute 'customer_clustering.py' in the terminal to train and dump the model objects first!")
    st.stop()

# Dashboard Main Header
st.title(" Enterprise Customer Analytics & Segmentation Hub")
st.markdown("Welcome to the AI-powered marketing strategy dashboard. Analyze overall customer base dynamics or predict new instances.")

# Setup tabs for modular navigation
tab1, tab2, tab3 = st.tabs([" Real-time Inference", " Customer Base Analytics", "📋 Raw Dataset Viewer"])

# --- TAB 1: Real-time User Prediction ---
with tab1:
    st.subheader("Predict Customer Segment")
    col1, col2 = st.columns(2)
    
    with col1:
        income = st.number_input("Annual Income (in $k):", min_value=15, max_value=150, value=50, step=1)
        spending_score = st.slider("Spending Score Indicator (1-100):", min_value=1, max_value=100, value=50)
    
    with col2:
        st.write(" ")
        st.write(" ")
        if st.button("Process Analytics Pipeline", use_container_width=True):
            user_payload = np.array([[income, spending_score]])
            user_payload_scaled = scaler.transform(user_payload)
            predicted_cluster = model.predict(user_payload_scaled)[0]
            
            # Map marketing business strategies to assigned clusters
            cluster_strategies = {
                0: ("Standard Core Customers", "Balanced income and spending pattern. Targeted with contextual newsletter updates and milestone discounts."),
                1: ("High-Value Whales (VIP)", "High spending, high income. Primary revenue catalyst. Retain with exclusive rewards, white-glove support, and early product previews."),
                2: ("Conservative Affluents", "High financial capacity, low consumption behavior. Engage using personalized premium offerings, high-quality bundles, and value justification campaigns."),
                3: ("High-Velocity Spenders", "Socio-economically lower-income bracket but extraordinarily reactive to trends. Target via flash sales, deferred payment options (BNPL), and urgency marketing."),
                4: ("Low-Priority Tier", "Minimal income and low engagement score. Low operational overhead allocation recommended. Standard automated drip campaigns only.")
            }
            
            segment_name, business_strategy = cluster_strategies.get(predicted_cluster, ("Standard", ""))
            
            st.success(f" Execution Success! Segment Assigned: **{segment_name}** (Cluster Index {predicted_cluster})")
            st.warning(f" **Operational Marketing Strategy:** {business_strategy}")

# --- TAB 2: Dynamic Visual Analytics ---
with tab2:
    st.subheader("Data Visualization & Model Performance Metrics")
    metric_col1, metric_col2, metric_col3 = st.columns(3)
    metric_col1.metric("Total Sample Size", f"{len(raw_df)} Customers")
    metric_col2.metric("Configured Clusters (K)", "5 Segments")
    metric_col3.metric("Clustering Engine", "Scikit-Learn KMeans")
    
    st.markdown("---")
    viz_col1, viz_col2 = st.columns(2)
    
    with viz_col1:
        st.markdown("### Model Segment Mapping")
        fig, ax = plt.subplots(figsize=(10, 6))
        sns.scatterplot(
            data=raw_df, 
            x='Annual Income (k$)', 
            y='Spending Score (1-100)', 
            hue='Cluster', 
            palette='tab10', 
            s=100, 
            style='Cluster',
            ax=ax
        )
        plt.title('Customer Segments Distribution')
        st.pyplot(fig)
            
    with viz_col2:
        st.markdown("### Mathematical Verification (Elbow Curve)")
        wcss = []
        for i in range(1, 11):
            from sklearn.cluster import KMeans
            kmeans_check = KMeans(n_clusters=i, init='k-means++', random_state=42)
            kmeans_check.fit(X_scaled)
            wcss.append(kmeans_check.inertia_)
            
        fig, ax = plt.subplots(figsize=(8, 4))
        plt.plot(range(1, 11), wcss, marker='o', linestyle='--', color='b')
        plt.title('The Elbow Point Graph (Model Evaluation)')
        plt.xlabel('Number of Clusters')
        plt.ylabel('WCSS')
        st.pyplot(fig)

# --- TAB 3: Data Table Index ---
with tab3:
    st.subheader("Segmented Database Index")
    st.dataframe(raw_df, use_container_width=True)
    
    st.markdown("### Segment Averages Table")
    summary_table = raw_df.groupby('Cluster').agg({
        'Age': 'mean',
        'Annual Income (k$)': 'mean',
        'Spending Score (1-100)': 'mean',
        'CustomerID': 'count'
    }).rename(columns={'CustomerID': 'Total Population Count'})
    st.table(summary_table)