src/streamlit_app.py

# Streamlit uygulaması

import streamlit as st
import pandas as pd
import pickle

st.set_page_config(
    page_title="Stellar Class Prediction",
    page_icon="🌌",
    layout="centered"
)

@st.cache_resource
def load_files():
    with open("src/stellar_class_model.pkl", "rb") as f:
        model = pickle.load(f)

    with open("src/target_encoder.pkl", "rb") as f:
        target_encoder = pickle.load(f)

    with open("src/feature_columns.pkl", "rb") as f:
        feature_columns = pickle.load(f)

    return model, target_encoder, feature_columns


model, target_encoder, feature_columns = load_files()

st.title("🌌 Stellar Class Prediction")
st.write(
    "Bu uygulama, astronomik gözlem verilerini kullanarak bir gök cisminin "
    "GALAXY, STAR veya QSO sınıfına ait olup olmadığını tahmin eder."
)

st.subheader("Gözlem Bilgilerini Giriniz")

alpha = st.number_input("Alpha", value=180.0)
delta = st.number_input("Delta", value=0.0)
u = st.number_input("u Bandı", value=22.0)
g = st.number_input("g Bandı", value=21.0)
r = st.number_input("r Bandı", value=20.0)
i = st.number_input("i Bandı", value=19.5)
z = st.number_input("z Bandı", value=19.0)
redshift = st.number_input("Redshift", value=0.5)

spectral_type = st.selectbox(
    "Spectral Type",
    ["M", "A/F", "G/K", "O/B"]
)

galaxy_population = st.selectbox(
    "Galaxy Population",
    ["Red_Sequence", "Blue_Cloud"]
)

if st.button("Tahmin Et"):
    input_data = pd.DataFrame({
        "alpha": [alpha],
        "delta": [delta],
        "u": [u],
        "g": [g],
        "r": [r],
        "i": [i],
        "z": [z],
        "redshift": [redshift],
        "spectral_type": [spectral_type],
        "galaxy_population": [galaxy_population]
    })

    input_data["u_g"] = input_data["u"] - input_data["g"]
    input_data["g_r"] = input_data["g"] - input_data["r"]
    input_data["r_i"] = input_data["r"] - input_data["i"]
    input_data["i_z"] = input_data["i"] - input_data["z"]

    input_data["total_brightness"] = (
        input_data["u"] +
        input_data["g"] +
        input_data["r"] +
        input_data["i"] +
        input_data["z"]
    )

    input_data["mean_brightness"] = input_data["total_brightness"] / 5

    input_data = pd.get_dummies(
        input_data,
        columns=["spectral_type", "galaxy_population"],
        drop_first=True
    )

    input_data = input_data.reindex(
        columns=feature_columns,
        fill_value=0
    )

    prediction = model.predict(input_data)
    prediction_label = target_encoder.inverse_transform(
        prediction.astype(int).ravel()
    )[0]

    st.success(f"Tahmin Edilen Sınıf: {prediction_label}")

    st.write("Girilen verilerden oluşturulan özellikler:")
    st.dataframe(input_data)