app.py

from datetime import datetime
from joblib import load
from PIL import Image

import pandas as pd
import random
import streamlit as st

model = load("model_title_update.joblib")

session = st.session_state

# if 'df' not in session:
#    session.df = pd.read_csv("model_log.csv").drop_duplicates()

if 'user_id' not in session:
    session.user_id = random.randint(1, 1_000_000)

st.markdown('# Titanic Workshop - Would you be able to survive the Titanic disaster?')
image = Image.open('titanic.jpeg')
st.image(image)
st.markdown("## ")
st.markdown("### Input your passenger data: ")

age = st.number_input('Age', min_value=1, max_value=80)

sex = st.radio("Sex", ("Male", "Female"))

name = st.text_input("Name", "")

if sex == "Male":
    title = st.radio("Title", ("Doctor", "Master", "Reverend"))
else:
    title = st.radio("Title", ("Doctor", "Master", "Miss", "Mrs", "Reverend"))

embark_port = st.radio(
    "Port of Embarkation",
    ('Cherbourg', 'Queenstown', 'Southampton'))

p_class = st.radio(
    "Passenger Class",
    ('Class 1', 'Class 2', 'Class 3'))

siblings = st.number_input('Number of Spouses Aboard', min_value=0, max_value=5)

parents = st.number_input("Number of Children Aboard", min_value=0, max_value=6)


passenger_data = {
    'Age': age,
    'is_male': 1 if sex == "Male" else 0,
    'Pclass_class 1': 1 if p_class == "Class 1" else 0,
    'Pclass_class 2': 1 if p_class == "Class 2" else 0,
    'Pclass_class 3': 1 if p_class == "Class 3" else 0,
    'Embarked_C': 1 if embark_port[0] == "C" else 0,
    'Embarked_Q': 1 if embark_port[0] == "Q" else 0,
    'Embarked_S': 1 if embark_port[0] == "S" else 0,
    'SibSp': siblings,
    'Parch': parents,
    'Mr': 1 if title == "Mr" else 0,
    'Miss': 1 if title == "Miss" else 0,
    'Mrs': 1 if title == "Mrs" else 0,
    "Master": 1 if title == "Master" else 0,
    'Dr': 1 if title == "Doctor" else 0,
    "Rev": 1 if title == "Reverend" else 0,
    'Other': 0,
    "name_len": len(name),
}

if st.button("Calculate my chances of survival"):
    passenger_data_df = pd.DataFrame.from_dict([passenger_data])
    prediction = model.predict(passenger_data_df)[0]
    prediction_proba = model.predict_proba(passenger_data_df)[:,1][0]

    if prediction == 0:
        st.markdown("# YOU'RE DEAD :skull:")
        
    elif sex == "Male":
        st.markdown("# YOU MADE IT, :man-swimming:")
    
    else:
        st.markdown("# YOU MADE IT, :woman-swimming:")
        
    
    st.write(f"Probabilty of survival: {round(prediction_proba, 2)* 100}%.")

    # passenger_data['user_id'] = session.user_id
    # passenger_data['date'] = str(datetime.now())
    # passenger_data["survived"] = prediction
    # passenger_data["survival_prob"] = prediction_proba
    
    # session.df = pd.concat(
    #    [
    #        session.df,
    #        pd.DataFrame([passenger_data])
    #    ],
    #    ignore_index=True
    #    )
    
    # session.df.drop_duplicates().to_csv('model_log.csv', mode="a", header=False, index=False)

# session.df = pd.read_csv("model_log.csv")
# session.df = session.df.drop_duplicates()
# session.df.to_csv("model_log.csv", index=False)