streamlit_app.py

import altair as alt
import pandas as pd
import streamlit as st

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import wget
import sklearn
import joblib

from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score


# Configuration Page
st.set_page_config(
    page_title="Classification titanic", page_icon="🤖", layout="centered"
)

# Titre de l'app
st.title("Classification binaire du titanic - 2023")


# st.markdown(
#     "[![Foo](https://upload.wikimedia.org/wikipedia/en/1/18/Titanic_%281997_film%29_poster.png)](http://google.com.au/)"
# )

st.markdown(
    '<div style="text-align: center;"><img src="https://upload.wikimedia.org/wikipedia/en/1/18/Titanic_%281997_film%29_poster.png" alt="Italian Trulli"></div>',
    unsafe_allow_html=True,
)

st.markdown("")
st.markdown("")

#### IMPORTATION DES DONNÉES #####
# wget.download(
#     "https://raw.githubusercontent.com/iid-ulaval/EEAA-datasets/master/titanic_train.csv",
#     "./titanic_train.csv",
# )
# wget.download(
#     "https://raw.githubusercontent.com/iid-ulaval/EEAA-datasets/master/titanic_test.csv",
#     "./titanic_test.csv",
# )
# train_data = pd.read_csv("titanic_train.csv")
# test_data = pd.read_csv("titanic_test.csv")

#### TEST VIZ ######
# st.dataframe(train_data.head(20))

# fig = plt.figure(figsize=(10, 4))
# sns.barplot(x="Pclass", y="Survived", data=train_data)
# st.pyplot(fig)
#### TEST VIZ ######


# # Traitement valeur manquantes
# train_data = train_data.dropna()

# # Traitement de la variable Sexe
# train_data["Sex"] = train_data["Sex"].replace("male", 1)
# train_data["Sex"] = train_data["Sex"].replace("female", 0)

# # EMBARKED
# train_data["Embarked"] = train_data["Embarked"].replace("C", 0)
# train_data["Embarked"] = train_data["Embarked"].replace("S", 1)
# train_data["Embarked"] = train_data["Embarked"].replace("Q", 2)

# # Ici on sépare nos données X (variables prédictives) et y (variables à prédire)
# X = train_data[
#     ["Sex", "Age", "Pclass", "Embarked"]
# ]  # variables prédictives (indépendantes)
# y = train_data["Survived"]  # Variable à prédire (dépendantes)

# model = LogisticRegression()  # Importe l'algorithme
# model.fit(X, y)


# load the saved model
model = joblib.load('model_titanic.joblib')

with st.form("my_form"):

    AGE = st.slider("Age de la personne?", 0, 2, 95)

    st.markdown("")
    st.markdown("")

    SEX = st.radio("Sexe de la personne", ("Homme", "Femme"))

    st.markdown("")
    st.markdown("")

    PCLASS = st.selectbox(
        "Séletionez la classe de la personne", ("Première", "Deuxième", "Troisème")
    )

    # EMBARKED
    # EMBARKED = st.selectbox("Séletionez l'embarcation", ("C", "S", "Q"))

    st.markdown("")
    st.markdown("")
    st.markdown("")
    st.markdown("")
    st.markdown("")
    st.markdown("")

    st.write(
        "Cette personne avait ",
        AGE,
        "ans,",
        " était un/une",
        SEX,
        "et était dans la",
        PCLASS,
        "classe",
    )

    if SEX == "Homme":
        SEX = 1
    else:
        SEX = 0

    if PCLASS == "Première":
        PCLASS = 1
    elif PCLASS == "Deuxième":
        PCLASS = 2
    else:
        PCLASS = 3

    # if EMBARKED == "C":
    #     EMBARKED = 1
    # elif EMBARKED == "S":
    #     EMBARKED = 2
    # else:
    #     EMBARKED = 3

    # PREDICTIONS 0 ou 1
    pred = model.predict(
        [[PCLASS, SEX, AGE]]
    )

    if pred == 0:
        pred = "mort"
    else:
        pred = "survie"

    st.metric(" ", pred)
    
    proba = model.predict_proba([[PCLASS, SEX, AGE]])

    st.write(f"Probabilité de survie : {proba[0][1]*100:.2f}%")

    st.write(f"Probabilité de décès : {proba[0][0]*100:.2f}%")

    submitted = st.form_submit_button("Prédire")