app2.py

def app(): #individual test
    import streamlit as st
    import pandas as pd
    from sklearn.model_selection import train_test_split
    from sklearn.linear_model import LogisticRegression
    from sklearn import metrics
    import numpy as np

    #title and stuff idk
    st.title('Credit Loan Risk')
    st.subheader('Will you get approved for a loan?')


    # data and model loading

    #loading data - updated already
    path = 'https://raw.githubusercontent.com/seescoto/cds303_model_app/main/updated_csv.csv'
    credit_risk = pd.read_csv(path)
    Feature_cols = ['interest_rate', 'loan_percent_income']


    #fitting the model
    #setting x and y
    X = credit_risk[Feature_cols]
    y = credit_risk.loan_status
    #split into test/train sets - no randomness so same results each time
    X_train,X_test,y_train,y_test = train_test_split(X, y, test_size=0.20, random_state = 0)
    #creating model and fitting it
    logreg = LogisticRegression()
    logreg.fit(X_train, y_train)
    #y_pred = logreg.predict(X_test)

    #model and data fitted!


    #describe interest rate stuff for looking at it idk
    #st.text(credit_risk['interest_rate'].describe())


    ###use inputs in sidebar

    #loan amount
    #st.text()

    loan_amt = st.number_input('How much money would you like to loan in dollars?', value = 10000)
    if loan_amt <= 0:
        st.warning('Loan amount must be a positive non-zero value.')

    #income
    income = st.number_input('What is your annual income in dollars?', value = 50000)
    if income < 0:
        st.warning('Income can not be a negative value.')

    #loan percent interest calculated from above
    loan_percent_test = loan_amt/(income + 0.01) #so no dividing by 0 even if income is 0

    #interest rate
    opts = ['Yes, I know my interest rate',
    'No, please estimate for me']

    ints = st.selectbox(
    'Do you know what interest rate you would be approved for if given a loan?', opts)

    if ints == opts[0]: #if they know their interest rate
        int_test = st.number_input('Please enter the interest rate you were previously approved for:',
        value = 12.00)
    elif ints == opts[1]: #if they dont know
        #credit score to estimate interest rate
        #the bank gives an interest rate and a random person might not know what theirs would be
        #so we'll just give them one based off an educated guess
        credit_test = st.number_input('What is your current credit score?',
        value = 600)
        if credit_test < 0 or credit_test > 850:
            st.warning('Credit score must be an integer between 0 and 850.')
        int_rates = (list(np.linspace(20.58, 17.12, 500 - 300 + 1)) +
        list(np.linspace(17.11, 10.5, 600 - 501 + 1)) +
        list(np.linspace(10.49, 5.5, 660 - 601 + 1)) +
        list(np.linspace(5.49, 3.67, 780 - 661 + 1)) +
        list(np.linspace(3.66, 3.66, 850 - 781 + 1)))
                    #avg interest rates based on credit score
                    #https://www.businessinsider.com/personal-finance/average-auto-loan-interest-rate
                    #closest to credit_risk['interest_rate'].describe() stats
        #851 values for interest, one for each credit score 300 to 850
        #input for model - int_test
        if credit_test < 300:
            #if less than 300 credit score, max interest
            int_test = 36.0
        else:
            int_test = int_rates[credit_test - 300]
        st.caption(f'Based on your credit score, your interest rate should be around {round(int_test, 2)}%')
        st.caption('Estimations of interest rate based on credit score may not yield accurate results.')
    else:
        st.warning(f'Please enter either your interest rate or credit score')




    ##modeling and getting the answer after checkbox has been selected
    input = [[int_test, loan_percent_test]]
    pred = logreg.predict(input) #predict defaulting [1] - aka NOT paying. 0 is approved/not default
    prob = logreg.predict_proba(input) #predict possibility of not defaulting and not defaulting
    if pred == 0:
        st.text(f'Our model predicts you will get approved for a loan!')
        st.text(f'The probability that you will be approved is {round(prob[0][0]*100, 2)}%')
    else:
        st.text(f'Our model predicts that you will not get approved for a loan.')
        st.text(f'The probability that you will be approved is {round(prob[0][0]*100, 2)}%')




    st.header("For bugs encountered, please contact us!")
    contact_form = """
    <form action="https://formsubmit.co/davis1kajuna@gmail.com" method="POST">
     <input type="hidden" name="_captcha" value="false">
     <input type="text" name="name" placeholder="Your name" required>
     <input type="email" name="email" placeholder="Your email" required>
     <textarea name="message" placeholder="Your message here."></textarea>
     <button type="submit">Send</button>
    </form>
    """

    st.markdown(contact_form, unsafe_allow_html=True)

    # Use Local CSS File
    def local_css(file_name):
        with open(file_name) as f:
            st.markdown(f"<style>{f.read()}</style>", unsafe_allow_html=True)


    css_path = "style/style.css"
    local_css(css_path)