eda.py

# import libraries
import streamlit as st
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

def app():
    st.title('Exploratory Data Analysis')

    df = pd.read_csv('eda_data.csv')
    st.subheader('Dataset Preview')
    st.write(df)

    st.subheader('Data Analysis Questions')
    st.write('How is the percentage of default payment as education level increases?')
    vis_1(df)

    st.write('How is the contribution of each gender to default payment?')
    vis_2(df)

    st.write('Which one got more into default payment, customers with limit balance above or below average?')
    vis_3(df)

    st.write('How does the average of default payment changes as the total late payment rises?')
    vis_4(df)

    st.write('How is the contribution of each marital status to default payment?')
    vis_5(df)


def vis_1(df):
    # make a copy of df
    df_eda_1 = df.copy()

    # get default payment sum of each education level
    df_eda_1_grouped_1 = pd.DataFrame(df_eda_1.groupby(by='education_level')['default_payment_next_month'].sum()).reset_index()

    # get row count of each education level
    df_eda_1_grouped_2 = pd.DataFrame(df_eda_1.groupby(by='education_level')['default_payment_next_month'].count()).reset_index()
    df_eda_1_grouped_2.rename(columns={'default_payment_next_month' : 'amount'}, inplace=True)

    # get default payment percentage of each educatin level
    dp_percentage_data = []

    for i in range(len(df_eda_1_grouped_1)):
        dp_percentage = df_eda_1_grouped_1['default_payment_next_month'][i] / df_eda_1_grouped_2['amount'][i]
        dp_percentage *= 100 
        dp_percentage = round(dp_percentage, 2)
        dp_percentage_data.append(dp_percentage)

    # create dataframe with education level and default payment percentage data
    df_eda_1_final = pd.DataFrame({
        'education_level' : df_eda_1_grouped_1['education_level'],
        'default_payment_percentage' : dp_percentage_data
    })

    # set plot's title
    plt.title('Default Payment Percentage for each Education Level')

    # define plot
    plt.bar(data=df_eda_1_final, x='education_level', height='default_payment_percentage')

    # set y limit
    ax = plt.gca()
    ax.set_ylim([0, 100])

    # add axis label
    ax.set_xlabel('education level')
    ax.set_ylabel('default payment percentage')

    # edit x ticks and their labels
    ax.set_xticks([1, 2, 3, 4])
    ax.set_xticklabels(['graduate_school', 'university', 'highschool', 'others'])

    # add bar label
    rects = ax.patches
    for rect in rects:
        height = rect.get_height()
        ax.text(rect.get_x() + rect.get_width()/2., 1.05*height,
                f'{height}%',
                ha='center', va='bottom')

    # show plot
    st.pyplot(plt.gcf())
    plt.clf()

def vis_2(df):
    # make a copy of df
    df_eda_2 = df.copy()

    # group df by sex and sum their default_payment
    df_eda_2_grouped = pd.DataFrame(df_eda_2.groupby(by='sex')['default_payment_next_month'].sum()).reset_index()

    # set plot's title
    plt.title('Gender Contribution to Default Payment')

    # define plot
    plt.pie(data=df_eda_2_grouped, x='default_payment_next_month', labels=['male', 'female'], autopct='%1.1f%%')

    # show plot
    st.pyplot(plt.gcf())
    plt.clf()

def vis_3(df):
    # make a copy of df
    df_eda_3 = df.copy()

    # get the average limit_balance
    avg_lim = df_eda_3['limit_balance'].mean()

    # get the limit group data, below or above average
    lim_group = []
    for lim in df_eda_3['limit_balance']:
        if lim < avg_lim:
            lim_group.append('below_average')
        else:
            lim_group.append('above_average')

    # add column
    df_eda_3['limit_group'] = lim_group

    # group df by limit group and sum their default payment count
    df_eda_3_grouped = pd.DataFrame(df_eda_3.groupby(by='limit_group')['default_payment_next_month'].sum()).reset_index()

    # set plot's title
    plt.title('Default Payment Amount Categorized by Limit Group')

    # define plot
    barh = plt.barh(data=df_eda_3_grouped, y='limit_group', width='default_payment_next_month', label='default_payment_next_month')

    # add axis label
    ax = plt.gca()
    ax.set_xlabel('default payment amount')
    ax.set_ylabel('limit balance group')

    # set x limit
    ax.set_xlim([0, 550])

    # add label
    ax = plt.gca()
    ax.bar_label(barh)

    # show plot
    st.pyplot(plt.gcf())
    plt.clf()

def vis_4(df):
    # make a copy of df
    df_eda_4 = df.copy()

    # get total late payment
    total_late_payment = pd.Series()
    for i in range(1, 7):
        total_late_payment = total_late_payment.add(df_eda_4[f'pay_{i}'], fill_value=0)

    # add column to df
    df_eda_4['total_late_payment'] = total_late_payment

    # group df by total late payment and sum default payment amount
    df_eda_4_grouped = pd.DataFrame(df_eda_4.groupby(by='total_late_payment')['default_payment_next_month'].mean()).reset_index()

    # set plot's title
    plt.title('The Effect of Late Payment to Default Payment')

    # define plot
    plt.plot(df_eda_4_grouped['total_late_payment'], df_eda_4_grouped['default_payment_next_month'])

    # add axis label
    ax = plt.gca()
    ax.set_xlabel('total late payment (month)')
    ax.set_ylabel('average default payment')

    # show plot
    st.pyplot(plt.gcf())
    plt.clf()

def vis_5(df):
    # make a copy of df
    df_eda_5 = df.copy()

    # group df by marital status and sum their default_payment
    df_eda_5_grouped = pd.DataFrame(df_eda_5.groupby(by='marital_status')['default_payment_next_month'].sum()).reset_index()

    # set plot's title
    plt.title('Contribution to Default Payment by Marital Status ')

    # define plot
    plt.pie(data=df_eda_5_grouped, x='default_payment_next_month', labels=['others', 'married', 'single', 'divorced'], autopct='%1.1f%%')

    # show plot
    st.pyplot(plt.gcf())
    plt.clf()