prediction.py

import streamlit as st
import pandas as pd
import numpy as np
import tensorflow 
from tensorflow.keras.models import load_model
import datetime
import pickle




# Load All Files

with open('final_pipeline.pkl', 'rb') as file_1:
  model_pipeline = pickle.load(file_1)

model_ann = load_model('churn_model.h5')



# bikin fungsi  
def run():

  with st.form(key='churn_data'):

      user_id = st.text_input('User ID', value='')
      age  = st.number_input('Age', min_value=10, max_value=70, value=25, help='Customer Age')
      gender  = st.selectbox('Gender', ('F','M'), index='M', help='M = Male  F= Female')
      region_category = st.selectbox('Region category', ('Town','Village','City'), index='City')
      membership_category = st.selectbox('Membership', ('No Membership','Basic Membership','Silver Membership', 'Gold Membership','Platinum Membership','Premium Membership'), index='Basic Membership')
      joining_date = st.date_input('Joining date',datetime.date(2019, 7, 6))
      joined_through_referral = st.selectbox('Join using referral', ('Yes','No'), index='No')
      preferred_offer_types = st.selectbox('preferred offer', ('Gift Vouchers/Coupons','Without Offers','Credit/Debit Card Offers'), index='Without Offers')
      medium_of_operation = st.selectbox('device ', ('Desktop','Smartphone','Both'), index='Smartphone')
      internet_option= st.selectbox('Internet', ('Mobile_data','Fiber_Optic','Wi-Fi'), index='Wi-Fi')
      days_since_last_login = st.number_input('How many days since last login', min_value=0, max_value=30, value=5)
      avg_time_spent = st.number_input('Avg time login', min_value=0, max_value=3000, value=5)
      avg_transaction_value = st.number_input('Avg transaction value', min_value=0, max_value=100000, value=1000)
      avg_frequency_login_days= st.number_input('Avg freq login', min_value=0, max_value=100, value=5)
      points_in_wallet= st.number_input('Avg time login', min_value=0, max_value=3000, value=5)
      used_special_discount = st.selectbox('Spesial discount', ('Yes','No'), index='No')
      offer_application_preference = st.selectbox('app preference', ('Yes','No'), index='No')
      past_complaint = st.selectbox('past complaint', ('Yes','No'), index='No')
      complaint_status  = st.selectbox('Complain status', ('No Information Available','Not Applicable','Solved','Solved in Follow-up','Unsolved'), index='Not Applicable')
      feedback  = st.selectbox('Complain status', ('User Friendly Website','Too many ads','Reasonable Price','Quality Customer Care','Products always in Stock','Poor Website','Poor Product Quality','Poor Customer Service'), index='No reason specified') 
           
      st.markdown('---')

      submitted = st.form_submit_button('Predict')

  data_inf = {
    'user_id':user_id,
    'age': age,
    'gender':gender,
    'region_category': region_category,
    'membership_category': membership_category,
    'joining_date': joining_date,
    'joined_through_referral':joined_through_referral,
    'preferred_offer_types': preferred_offer_types,
    'medium_of_operation': medium_of_operation,
    'internet_option': internet_option,
    'days_since_last_login': days_since_last_login,
    'avg_time_spent': avg_time_spent,
    'avg_transaction_value': avg_transaction_value,
    'avg_frequency_login_days': avg_frequency_login_days,
    'points_in_wallet': points_in_wallet,
    'used_special_discount': used_special_discount,
    'offer_application_preference': offer_application_preference,
    'past_complaint': past_complaint,
    'complaint_status': complaint_status,
    'feedback': feedback
  }

  data_inf = pd.DataFrame([data_inf])
  # Create Binning frequency login
  bins = [-1, 10, 20, 30, 40, 50, 100]
  labels =[1,2,3,4,6,7]
  data_inf['binned_frequency_login'] = pd.cut(data_inf['avg_frequency_login_days'], bins,labels=labels).astype(float)
  st.dataframe(data_inf)

  if submitted:
      
      # transform data inference
      data_inf_transform = model_pipeline.transform(data_inf)

      # Predict using model ann
      y_pred_inf = model_ann.predict(data_inf_transform)
      y_pred_inf = np.where(y_pred_inf >= 0.5, 1, 0)
      if y_pred_inf.any() == 1:
            st.write('## The Customer probably will CHURN')
      else:
            st.write('## The Customer probably will NOT Churn')

      
if __name__ == '__main__':
    run()