app.py

# app.py
import streamlit as st
import pandas as pd
import unzip
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import LabelEncoder, StandardScaler
from sklearn.decomposition import PCA
from sklearn.cluster import AgglomerativeClustering

st.set_page_config(layout="wide")

@st.cache_data
def load_data():
    data = pd.read_csv("dataset/marketing_campaign.csv", sep="\t")
    data = data.dropna()
    data["Dt_Customer"] = pd.to_datetime(data["Dt_Customer"], errors='coerce')
    data["Customer_For"] = (data["Dt_Customer"].max() - data["Dt_Customer"]).dt.days
    data["Age"] = 2025 - data["Year_Birth"]
    data["Spent"] = data[["MntWines","MntFruits","MntMeatProducts","MntFishProducts","MntSweetProducts","MntGoldProds"]].sum(axis=1)
    data["Living_With"]=data["Marital_Status"].replace({"Married":"Partner", "Together":"Partner", "Absurd":"Alone", "Widow":"Alone", "YOLO":"Alone", "Divorced":"Alone", "Single":"Alone",})
    data["Children"] = data["Kidhome"] + data["Teenhome"]
    data["Family_Size"] = data["Living_With"].replace({"Alone": 1, "Partner":2}) + data["Children"]
    data["Is_Parent"] = np.where(data["Children"]> 0, 1, 0)
    data["Education"] = data["Education"].replace({"Basic":"Undergraduate","2n Cycle":"Undergraduate", "Graduation":"Graduate", "Master":"Postgraduate", "PhD":"Postgraduate"})
    data = data.rename(columns={"MntWines": "Wines","MntFruits":"Fruits","MntMeatProducts":"Meat","MntFishProducts":"Fish","MntSweetProducts":"Sweets","MntGoldProds":"Gold"})
    to_drop = ["Marital_Status", "Dt_Customer", "Z_CostContact", "Z_Revenue", "Year_Birth", "ID"]
    data = data.drop(to_drop, axis=1)
    data = data[(data["Age"]<90) & (data["Income"]<600000)]

    # Label Encoding
    s = (data.dtypes == 'object')
    object_cols = list(s[s].index)
    for i in object_cols:
        data[i] = LabelEncoder().fit_transform(data[i])

    return data

data = load_data()

# Sidebar
st.sidebar.title("Customer Segmentation")
st.sidebar.write("Using Agglomerative Clustering + PCA")

st.sidebar.header("🔍 Predict Cluster for New Customer")

with st.sidebar.form("customer_form"):
    income = st.number_input("Income", min_value=0, max_value=300000, step=5000)
    recency = st.slider("Recency (days since last purchase)", 0, 100, 10)
    customer_for = st.slider("Customer For (days since enrolled)", 0, 3650, 1000)
    age = st.slider("Age", 18, 100, 40)
    children = st.slider("Children (kids + teens)", 0, 5, 0)
    is_parent = st.selectbox("Is Parent?", ["No", "Yes"])
    education = st.selectbox("Education", ["Undergraduate", "Graduate", "Postgraduate"])
    living_with = st.selectbox("Living With", ["Alone", "Partner"])
    spent = st.number_input("Total Amount Spent", min_value=0, max_value=10000, step=100)

    submitted = st.form_submit_button("Predict Cluster")
# PCA and clustering
features = data.drop(columns=['AcceptedCmp1','AcceptedCmp2','AcceptedCmp3','AcceptedCmp4','AcceptedCmp5','Complain','Response'])
# Drop NaNs in both features and data
valid_index = features.dropna().index
features = features.loc[valid_index]
data = data.loc[valid_index]

# Proceed with scaling + PCA
scaled = StandardScaler().fit_transform(features)
pca = PCA(n_components=3)
pca_data = pca.fit_transform(scaled)
ac = AgglomerativeClustering(n_clusters=4)
clusters = ac.fit_predict(pca_data)



data["Cluster"] = clusters

if submitted:
    # Encode categorical fields
    is_parent_val = 1 if is_parent == "Yes" else 0
    education_val = {"Undergraduate": 0, "Graduate": 1, "Postgraduate": 2}[education]
    living_val = {"Alone": 0, "Partner": 1}[living_with]
    family_size = (1 if living_with == "Alone" else 2) + children

    # Create new record
    user_row = {
        'Income': income,
        'Recency': recency,
        'Customer_For': customer_for,
        'Age': age,
        'Children': children,
        'Family_Size': family_size,
        'Is_Parent': is_parent_val,
        'Education': education_val,
        'Living_With': living_val,
        'Spent': spent
    }

    # Use same columns as features used earlier
    input_df = pd.DataFrame([user_row])
    combined = pd.concat([features, input_df], ignore_index=True)
    combined = combined.dropna()

    # Scale + PCA
    scaled_combined = StandardScaler().fit_transform(combined)
    pca_combined = PCA(n_components=3).fit_transform(scaled_combined)

    # Run clustering again
    predicted_clusters = AgglomerativeClustering(n_clusters=4).fit_predict(pca_combined)
    predicted_label = predicted_clusters[-1]

    st.sidebar.success(f"Predicted Cluster for User: 🎯 **Cluster {predicted_label}**")
    cluster_messages = {
    0: "🧠 High-value cluster: Loyal customers!",
    1: "💼 Moderate income & engagement. Upsell possible.",
    2: "🔍 Low spenders, high recency. Need attention.",
    3: "🎯 Deal-responsive group. Good for promos."
    }
    # Optional: Show user PCA location
    fig = plt.figure(figsize=(6, 5))
    ax = fig.add_subplot(111, projection='3d')
    ax.scatter(pca_combined[:-1, 0], pca_combined[:-1, 1], pca_combined[:-1, 2], c=predicted_clusters[:-1], cmap='tab10', label='Existing')
    ax.scatter(pca_combined[-1, 0], pca_combined[-1, 1], pca_combined[-1, 2], c='black', s=100, label='You')
    ax.set_title("PCA + Cluster Visualization")
    ax.legend()
    st.sidebar.pyplot(fig)

    st.sidebar.info(cluster_messages.get(predicted_label, "Unknown cluster"))
    
st.subheader("Cluster-wise Spending vs Income")
import seaborn as sns
import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(10, 6))
sns.scatterplot(data=data, x="Income", y="Spent", hue="Cluster", palette="tab10", ax=ax)
st.pyplot(fig)

st.subheader("Cluster Distribution")
st.bar_chart(data["Cluster"].value_counts())

st.subheader("View Clustered Data")
st.dataframe(data)