eda.py

import streamlit as st
import pandas as pd 
import matplotlib.pyplot as plt 

# This cell is used to make new column called 'group_age'
def filter_group_age(age):
    if 0 <= age <= 14:
        return "children"
    elif 15 <= age <= 24:
        return "youth"
    elif 25 <= age <= 64: 
        return "adults"
    elif age >= 65:
        return "seniors"
    else: 
        return "unknown"



def run():
    st.title('Credit Card Payment Exploration Data Analyst(EDA)')
    st.image('https://www.i1.creditdonkey.com/image/1/[email protected]')
    st.write('This page is made by Yudis Aditya')
    st.markdown('---')

    st.write('In this page, I want to show visualization data based on my analyst process about credit card payment. This EDA has information about customer segmentation and customer behaviour')
    st.link_button('Link dataset','https://console.cloud.google.com/bigquery?p=bigquery-public-data&d=ml_datasets&t=credit_card_default&page=table&project=direct-keel-424102-f5&ws=!1m5!1m4!4m3!1sbigquery-public-data!2sml_datasets!3scredit_card_default')


    df = pd.read_csv('P1G5_Set_1_yudis_aditya.csv')
    df['group_age'] = df['age'].apply(filter_group_age)

    st.write('### Distribution Data Age')
    # This cell is used to create histogram column age
    data = df['age']
    fig = plt.figure(figsize=(15,5))
    plt.hist(x=data)
    plt.title("Distribution Data Age")
    plt.show()
    st.pyplot(plt)
    st.write("From Graph 'Distribution Data Age' we can see that the most client who use credit card have age around 20 - 40")
    

    st.write('Based on Standarization Statistic Canada, we can make group of age like this :')
    st.write('- 0 - 14 = children ')
    st.write('- 15 - 24 = youth')
    st.write('- 25 - 64 = adults ')
    st.write('- more than equal 65 = seniors')
    
    
    data = df.groupby('group_age').size()
    x = data.index
    y = data.values
    fig = plt.figure(figsize=(15,5))
    plt.pie(y,labels=x,autopct='%1.1f%%')
    plt.title("Pie Chart Data based on Group Age")
    st.pyplot(plt)
    
    st.write('From Graph "Pie Chart Data based on Group Age" we can know that majority significant people who use credit card is in group age adults. and The second is youth but not really significant. And group age "seniors" is rare to use credit card')
   

    st.write('### Distribution Data Marital Status')
    # This cell is used to create pie graph to know distibution data based on marital_status
    data = df[df['marital_status'] != 0].groupby('marital_status').size()
    x = ['married','single','others']
    y = data.values

    fig = plt.figure(figsize=(15,5))
    plt.pie(y,labels=x,autopct='%1.1f%%')
    plt.title("Distribution data person based on marital status")
    st.pyplot(fig)

    st.write('From Graph above we can see that person who single is more to use credit card than married. person who has marital_status others has not really significant')
    
    st.write('### Distribution Data Gender')
    data = df.groupby('sex').size()
    x = ['male','female']
    y = data.values


    fig = plt.figure(figsize=(15,5))
    plt.pie(y,labels=x,autopct='%1.1f%%')
    plt.title("Distribution data person based on gender")
    st.pyplot(fig)

    st.write('From Graph above we can see that female is more using credit card than male')
    st.write('### Total People who pay duly on April - September in 2005')
    fig = plt.figure(figsize=(15,5))
    # This cell is used to create line chart to know total status payment duly in  2005
    data_6 = df[(df['pay_6'] <= 0)][['pay_6']].count()
    data_5 = df[(df['pay_5'] <= 0)][['pay_5']].count()
    data_4 = df[(df['pay_4'] <= 0)][['pay_4']].count()
    data_3 = df[(df['pay_3'] <= 0)][['pay_3']].count()
    data_2 = df[(df['pay_2'] <= 0)][['pay_2']].count()
    data_1 = df[(df['pay_0'] <= 0)][['pay_0']].count()

    data_input = {
        'Apr':data_1.values[0],
        'May':data_2.values[0],
        'Jun':data_3.values[0],
        'July':data_4.values[0],
        'August':data_5.values[0],
        'Sep':data_6.values[0],
    }

    data_visualize = pd.DataFrame([data_input])
    data_visualize.values.tolist()[0]

    # print(data)
    x = data_visualize.columns.to_list()
    y = data_visualize.values.tolist()[0]

    plt.plot(x,y)
    plt.title("Total status payment duly in 2005")
    st.pyplot(fig)
    st.write('From Graph above we can know that Total people who pay duly are increased significant from April 2005 until September 2005. ')
    
    st.write('### Median Bill Amount on April - September in 2005')
    fig = plt.figure(figsize=(15,5))
    # This cell is used to create line chart to know total status payment duly in  2005
    data_6 = df[['bill_amt_6']].median()
    data_5 = df[['bill_amt_5']].median()
    data_4 = df[['bill_amt_4']].median()
    data_3 = df[['bill_amt_3']].median()
    data_2 = df[['bill_amt_2']].median()
    data_1 = df[['bill_amt_1']].median()

    data_input = {
        'Apr':data_1.values[0],
        'May':data_2.values[0],
        'Jun':data_3.values[0],
        'July':data_4.values[0],
        'August':data_5.values[0],
        'Sep':data_6.values[0],
    }

    data_visualize = pd.DataFrame([data_input])
    data_visualize.values.tolist()[0]

    # print(data)
    x = data_visualize.columns.to_list()
    y = data_visualize.values.tolist()[0]

    plt.bar(x,y)
    plt.title("Median Bill amount from April - September 2005")
    st.pyplot(fig)
    st.write('From graph above i use median instead mean because distribution data is not normal. We can see that median bill amount from april until September is decrased. That visualize that most people has decrased bill amount every month, which is good.')
    
    st.write('### Total people who pay and not pay for next month (October) 2005')
    fig = plt.figure(figsize=(15,5))
    data_input = df.groupby(['default_payment_next_month']).size()

    x = ['Pay','Not Pay']
    y = data_input.values

    plt.bar(x,y)
    plt.title('Total people who pay and dont pay on October 2005')
    st.pyplot(fig)
    st.write('From Graph above we know that more people pay for next month (October) than not pay. It is good result because that indicate client good behaviour')
    
if __name__ == '__main__':
    run()