leyuzak/Binary-Classification-with-a-Bank-Dataset

1import joblib2import pandas as pd3import streamlit as st4 5# --- sklearn pickle uyumluluğu (sadece bu kalsın) ---6try:7    import sklearn.compose._column_transformer as _ct8    if not hasattr(_ct, "_RemainderColsList"):9        class _RemainderColsList(list):10            pass11        _ct._RemainderColsList = _RemainderColsList12except Exception:13    pass14# ---------------------------------------------------15 16st.set_page_config(17    page_title="Bank Subscription Predictor",18    page_icon="🏦",19    layout="centered"20)21 22@st.cache_resource23def load_model():24    return joblib.load("bank_subscription_model.joblib")25 26model = load_model()27 28st.title("🏦 Bank Subscription Predictor")29st.caption("Predict the probability that a client subscribes to a term deposit (y=1).")30 31with st.expander("ℹ️ What is this?"):32    st.markdown(33        """34This app loads a trained model (`bank_subscription_model.joblib`)35and predicts the probability of `y=1`.36Fill the inputs and click **Predict**.37"""38    )39 40st.subheader("Client Features")41 42col1, col2 = st.columns(2)43 44with col1:45    age = st.number_input("age", min_value=0, max_value=120, value=35, step=1)46    job = st.selectbox(47        "job",48        [49            "admin.", "blue-collar", "entrepreneur", "housemaid", "management",50            "retired", "self-employed", "services", "student", "technician",51            "unemployed", "unknown"52        ],53        index=454    )55    marital = st.selectbox("marital", ["divorced", "married", "single", "unknown"], index=1)56    education = st.selectbox(57        "education",58        [59            "basic.4y", "basic.6y", "basic.9y", "high.school", "illiterate",60            "professional.course", "university.degree", "unknown"61        ],62        index=663    )64    default = st.selectbox("default", ["no", "yes", "unknown"], index=0)65    balance = st.number_input("balance", value=0.0, step=10.0)66    housing = st.selectbox("housing", ["no", "yes", "unknown"], index=1)67    loan = st.selectbox("loan", ["no", "yes", "unknown"], index=0)68 69with col2:70    contact = st.selectbox("contact", ["cellular", "telephone", "unknown"], index=0)71    day = st.number_input("day", min_value=1, max_value=31, value=15, step=1)72    month = st.selectbox(73        "month",74        ["jan", "feb", "mar", "apr", "may", "jun", "jul", "aug", "sep", "oct", "nov", "dec"],75        index=476    )77    duration = st.number_input("duration", min_value=0, value=180, step=10)78    campaign = st.number_input("campaign", min_value=0, value=1, step=1)79    pdays = st.number_input("pdays", value=-1, step=1)80    previous = st.number_input("previous", min_value=0, value=0, step=1)81    poutcome = st.selectbox("poutcome", ["failure", "other", "success", "unknown"], index=3)82 83client_id_str = st.text_input("id (optional)", value="0")84 85try:86    client_id = int(client_id_str)87except Exception:88    client_id = 089 90x = pd.DataFrame([{91    "id": client_id,92    "age": age,93    "job": job,94    "marital": marital,95    "education": education,96    "default": default,97    "balance": float(balance),98    "housing": housing,99    "loan": loan,100    "contact": contact,101    "day": int(day),102    "month": month,103    "duration": int(duration),104    "campaign": int(campaign),105    "pdays": int(pdays),106    "previous": int(previous),107    "poutcome": poutcome108}])109 110st.divider()111 112if st.button("🔮 Predict", type="primary"):113    try:114        if hasattr(model, "predict_proba"):115            proba = float(model.predict_proba(x)[:, 1][0])116        else:117            score = float(model.decision_function(x)[0])118            import math119            proba = 1 / (1 + math.exp(-score))120 121        st.success(f"Predicted probability of y=1: **{proba:.4f}**")122 123        if proba >= 0.7:124            st.write("✅ High likelihood of subscription")125        elif proba >= 0.4:126            st.write("🟡 Medium likelihood of subscription")127        else:128            st.write("🔻 Low likelihood of subscription")129 130    except Exception as e:131        st.error(132            "Prediction failed. Check that the model file matches the expected feature columns."133        )134        st.exception(e)135