streamlit_app.py

# import libraries

import warnings
import streamlit as st
from streamlit_option_menu import option_menu
import pandas as pd
import seaborn as sns
import pandas as pd
import numpy as np
from sklearn import tree
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.pipeline import Pipeline
from sklearn.model_selection import GridSearchCV
from sklearn.decomposition import PCA
import plotly as px
import matplotlib.pyplot as plt
from pandas_profiling import ProfileReport
from streamlit_pandas_profiling import st_profile_report
from PIL import Image
from sklearn.model_selection import train_test_split
from sklearn.metrics import precision_score, recall_score, f1_score, confusion_matrix, accuracy_score

#  # Insert Pictures
# st.image("photo_2023.jpg")
# st.image("photo_2023.jpg")

# Insert A Navbar
with st.sidebar:
    selected = option_menu(
        menu_title='Menu',
        options=['Data Download', 'About'],
        icons=['file-arrow-down-fill', 'book'],
        menu_icon='hourglass-split',
        # default_index=0,
    )
if selected == 'Data Download':
    st.markdown("""
            1. Click the link below and right click raw button and tap **Save link as**.
            2. Select location and change the extention of file to .csv if the file is being downloaded as .txt.
            3. Upload the file in the sidebar to view pandas profiling report.
            [click here to download the heart attack dataset](https://github.com/ayazparhyar/streamlit_pkc_app/blob/main/heart.csv)"""
                )
elif selected == 'About':
    st.write("""
            #### Made with ❤️ by Qadir Shahbaz, Muhammad Ayaz, Muhammad Ali Farrukh & Muhammad Ali Butt :rocket:
            CopyRight © 2023 PKC Team1 All Rights Reserved.""")

# Insert Heading and Subheading
st.write("""
        # Heart Attack Analysis & Prediction :ambulance: """)

# Insert sidebar to upload dataset to the streamlit app CSV format
st.markdown("""
            # **01. Exploratory Data Analysis**""")
with st.sidebar.header("Upload your dataset(.csv)"):
    uploaded_file = st.sidebar.file_uploader("Upload your file", type=["csv"])
if uploaded_file is not None:
    @st.cache
    def load_csv():
        csv = pd.read_csv(uploaded_file)
        return csv
    df = load_csv()
    pr = ProfileReport(df, explorative=True)
    st.header("**Input DF**")
    st.write(df)
    st.write("---")
    st.header("**Profiling report with pandas**")
    st_profile_report(pr)
else:
    st.info("Awaiting for CSV file")
    if st.button("press to use example data"):
        def load_data():
            a = pd.DataFrame(np.random.rand(100, 5),
                             columns=["age", "banana", "Codenics", "duck", "Ear"])
            return a
        df1 = load_data()
        pr = ProfileReport(df, explorative=True)
        st.Header("**Input DF**")
        st.write(df)
        st.write("---")
        st.header("**Profiling report with pandas**")
        st_profile_report(pr)

# Insert User Input Parameters
st.sidebar.header("Patient Data")


def user_report():
    min_value = 25
    max_value = 80
    min_value_1 = 90
    max_value_1 = 200
    min_value_2 = 125
    max_value_2 = 565
    min_value_3 = 70
    max_value_3 = 202
    min_value_4 = 0.0
    max_value_4 = 7.0

    sex_values = ("0", "1")
    cp_values = ("0", "1", "2", "3")
    age = st.sidebar.slider("age", min_value, max_value,
                            value=None, step=None, format=None)
    sex = st.sidebar.selectbox("sex", sex_values)
    cp = st.sidebar.selectbox("Chest pain", cp_values)
    trtbps = st.sidebar.slider(
        "trtbps", min_value_1, max_value_1, value=None, step=None, format=None)
    chol = st.sidebar.slider(
        "chol", min_value_2, max_value_2, value=None, step=None, format=None)
    fbs = st.sidebar.selectbox("fbs", ("1", "2", "3"))
    restecg = st.sidebar.selectbox("restecg", ("0", "1"))
    thalachh = st.sidebar.slider(
        "thalachh", min_value_3, max_value_3, value=None, step=None, format=None)
    exng = st.sidebar.selectbox("exng", ("0", "1"))
    oldpeak = st.sidebar.slider(
        "oldpeak", min_value_4, max_value_4, value=None, step=None, format=None)
    slp = st.sidebar.selectbox("slp", ("0", "1", "2"))
    caa = st.sidebar.selectbox("caa", ("0", "1", "2", "3", "4"))
    thall = st.sidebar.selectbox("thall", ("0", "1", "2", "3"))

    user_report_data = {"age": age,
                        "sex": sex,
                        "cp": cp,
                        "trtbps": trtbps,
                        "chol": chol,
                        "fbs": fbs,
                        "restecg": restecg,
                        "thalachh": thalachh,
                        "exng": exng,
                        "oldpeak": oldpeak,
                        "slp": slp,
                        "caa": caa,
                        "thall": thall}
    report_data = pd.DataFrame(user_report_data, index=[0])
    return report_data


# Description of the Columns
st.write(""" ## Heart attack Analysis dataset's keys definition
         
**1. Age** : Age of the patient

**2. Sex** : Sex of the patient

    Value 0: Female
    
    Value 1: Male

**3. cp** : Chest Pain type

    Value 0: Typical angina

    Value 1: Atypical angina

    Value 2: Non-anginal pain

    Value 3: Asymptomatic

**4. trtbps** : Blood pressure after receiving treatment (in mm Hg)

**5. chol**: Cholesterol in mg/dl fetched via BMI sensor

**6. fbs**: (Fasting blood sugar > 120 mg/dl)

    1 = true

    0 = false

**7. rest_ecg**: Resting electrocardiographic results
    Value 0: normal

    Value 1: having ST-T wave abnormality (T wave inversions and/or ST elevation or depression of > 0.05 mV)

    Value 2: showing probable or definite left ventricular hypertrophy by Estes' criteria

**8. thalach**: Maximum heart rate achieved

**9.exang**: Exercise induced angina(discomfort du)

    1 = yes

    0 = no

**10. old peak**: ST depression induced by exercise relative to rest

**11. slp**: The slope of the peak exercise ST segment

    0 = Unsloping

    1 = flat

    2 = downsloping

**12. caa**: Number of major vessels (0-3)

**13. thall** : Thalassemia

    0 = null

    1 = fixed defect

    2 = normal

    3 = reversable defect

**14. output**: diagnosis of heart disease (angiographic disease status)

    0: < 50% diameter narrowing. less chance of heart disease

    1: > 50% diameter narrowing. more chance of heart disease""")

# To upload the file
df1 = pd.read_csv("heart.csv")

# Inserting another subheading
st.subheader("Heart Attack dataset")
st.write(df1)

# Performing EDA
st.subheader("List of Columns")
st.write(df1.columns)
st.subheader("Heart Attack dataset's description")
st.write(df1.describe().T)

# Removing duplicate value from dataset
df1.drop_duplicates(inplace=True)

# Patient data
st.title("Heart Attack prediction")
user_data = user_report()
st.subheader("Patient Data")

warnings.filterwarnings('ignore')

st.write("""
# Model Selection App
Select the best model by adjusting the hyperparameters!
""")

selected_model = st.selectbox(
    'Select a model', ['rf', 'knn', 'dt', 'lr'])


def get_params(selected_model):
    params = dict()
    if selected_model == 'rf':

        n_estimators = st.slider('n_estimators', 100, 20, step=1)
        criterion = st.selectbox('criterion', ['gini', 'entropy'])
        max_depth = st.slider('max_depth', 3, 100, step=1)
        params['n_estimators'] = n_estimators
        params['criterion'] = criterion
        params['max_depth'] = max_depth

    elif selected_model == 'knn':
        n_neighbors = st.slider('n_neighbors', 3, 100, step=1)
        weights = st.selectbox('weights', ['uniform', 'distance'])
        algorithm = st.selectbox(
            'algorithm', ['auto', 'ball_tree', 'kd_tree', 'brute'])
        params['n_neighbors'] = n_neighbors
        params['weights'] = weights
        params['algorithm'] = algorithm

    elif selected_model == 'dt':
        criterion = st.selectbox('criterion', ['gini', 'entropy'])
        max_depth = st.slider('max_depth', 3, 100, step=1)
        params['criterion'] = criterion
        params['max_depth'] = max_depth

    elif selected_model == 'lr':
        C = st.slider('C', 0.1, 1.0, step=0.1)
        penalty = st.selectbox('penalty', [None, 'l2'])
        params['C'] = C
        params['penalty'] = penalty
    return params


params = get_params(selected_model)


def get_classifier(selected_model, params):
    classifier = None
    if selected_model == 'rf':
        classifier = RandomForestClassifier(
            n_estimators=params['n_estimators'], criterion=params['criterion'], max_depth=params['max_depth'], random_state=1234)
    elif selected_model == 'knn':
        classifier = KNeighborsClassifier(
            n_neighbors=params['n_neighbors'], weights=params['weights'], algorithm=params['algorithm'])
    elif selected_model == 'dt':
        classifier = DecisionTreeClassifier(
            criterion=params['criterion'], max_depth=params['max_depth'], random_state=1234)
    elif selected_model == 'lr':
        classifier = LogisticRegression(
            C=params['C'], penalty=params['penalty'], random_state=1234)
    return classifier


model = get_classifier(selected_model, params)

# data splitting into X and y and Train test split
X = df1.drop(["output"], axis=1)
y = df1["output"]
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=1234)
model.fit(X_train, y_train)
user_result = model.predict(user_data)
st.write('Prediction', user_result)

acc = accuracy_score(y_test, model.predict(X_test))
st.write(f'Model = {selected_model}')
st.write(f'Parameters =', params)
st.write(f'Accuracy =', acc)

# output
st.header("Your Report:")
output = ''
if user_result[0] == 0:
    output = 'You are safe'
    st.balloons()
elif user_result[0] == 1:
    output = "Look after your health"
    # st.warning("Attack detected", "Please take necessary precautions.")
st.title(output)


# Visualizing the dataset

pca = PCA(2)
X_projected = pca.fit_transform(X)

# first principal component of the data 0 means first column
x1 = X_projected[:, 0]
# second principal component of the data 1 means second column
x2 = X_projected[:, 1]
fig = plt.figure()
plt.scatter(x1, x2, c=y, alpha=0.8, cmap='autumn', edgecolors='crimson', s=60)
plt.xlabel('Age', fontsize=15, color='red', fontweight='bold')
plt.ylabel('Sex', fontsize=15, color='red', fontweight='bold')
plt.colorbar()
st.pyplot(fig)
# Age vs Chol
st.header("Age vs Cholesterol")
fig_preg = plt.figure()
dimension1 = sns.scatterplot(
    x="age", y="chol", data=df1, hue="output", palette="autumn_r", s=60)
dimension2 = sns.scatterplot(
    x=user_data["age"], y=user_data["chol"], s=60)
plt.xticks(ha='right', rotation=45, fontsize=8)
plt.yticks(ha='right', rotation=45, fontsize=8)
plt.title("0 - Healty & 1 Possiblity of Heart attach",
          fontsize=15, color='crimson', fontweight='bold')
st.pyplot(fig_preg)