src/streamlit_app.py

import streamlit as st
import pandas as pd
import numpy as np
import pickle


st.title("Credit Card Customer Segmentation")

st.write(
    "Bu uygulama, KMeans kümeleme algoritmasını kullanarak müşteri segmentlerini tahmin eder."
)


# Model ve scaler yükleme
model = pickle.load(open("src/kmeans_model.pkl", "rb"))

scaler = pickle.load(open("src/scaler.pkl", "rb"))


# Kullanıcı girişleri
balance = st.number_input("Balance", value=1000.0)

balance_frequency = st.slider(
    "Balance Frequency",
    0.0,
    1.0,
    0.9
)

purchases = st.number_input("Purchases", value=500.0)

oneoff_purchases = st.number_input("Oneoff Purchases", value=200.0)

installments_purchases = st.number_input(
    "Installments Purchases",
    value=300.0
)

cash_advance = st.number_input("Cash Advance", value=0.0)

purchases_frequency = st.slider(
    "Purchases Frequency",
    0.0,
    1.0,
    0.5
)

oneoff_frequency = st.slider(
    "Oneoff Purchases Frequency",
    0.0,
    1.0,
    0.3
)

installments_frequency = st.slider(
    "Installments Purchases Frequency",
    0.0,
    1.0,
    0.4
)

credit_limit = st.number_input("Credit Limit", value=5000.0)

payments = st.number_input("Payments", value=1000.0)

minimum_payments = st.number_input(
    "Minimum Payments",
    value=300.0
)

full_payment = st.slider(
    "Full Payment Ratio",
    0.0,
    1.0,
    0.3
)

tenure = st.slider(
    "Tenure",
    1,
    12,
    12
)


# DataFrame oluşturma
input_data = pd.DataFrame({
    
    "BALANCE":[balance],

    "BALANCE_FREQUENCY":[balance_frequency],
    
    "PURCHASES":[purchases],
    
    "ONEOFF_PURCHASES":[oneoff_purchases],
    
    "INSTALLMENTS_PURCHASES":[installments_purchases],
    
    "CASH_ADVANCE":[cash_advance],
    
    "PURCHASES_FREQUENCY":[purchases_frequency],
    
    "ONEOFF_PURCHASES_FREQUENCY":[oneoff_frequency],
    
    "PURCHASES_INSTALLMENTS_FREQUENCY":[installments_frequency],
    
    "CASH_ADVANCE_FREQUENCY":[0],
    
    "CASH_ADVANCE_TRX":[0],
    
    "PURCHASES_TRX":[10],
    
    "CREDIT_LIMIT":[credit_limit],
    
    "PAYMENTS":[payments],
    
    "MINIMUM_PAYMENTS":[minimum_payments],
    
    "PRC_FULL_PAYMENT":[full_payment],
    
    "TENURE":[tenure]
})


# Log dönüşümü
input_data = np.log1p(input_data)


# Scaling
input_scaled = scaler.transform(input_data)


# Tahmin
if st.button("Predict Cluster"):

    cluster = model.predict(input_scaled)

    st.subheader(f"Predicted Cluster: {cluster[0]}")

    
    if cluster[0] == 0:
        st.write("Risky cash advance users")

    elif cluster[0] == 1:
        st.write("Premium high spending customers")

    elif cluster[0] == 2:
        st.write("Balanced active customers")

    elif cluster[0] == 3:
        st.write("Low spending passive customers")

    else:
        st.write("Low balance customers")