app.py

# import required libraries

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import pickle
import joblib
import os

from datetime import datetime
from datetime import timedelta
from sklearn.model_selection import RandomizedSearchCV, GridSearchCV, train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import r2_score
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import StandardScaler
import streamlit as st 
import warnings
warnings.filterwarnings('ignore')

st.title("Predict Unrolled Values")
st.sidebar.header('Enter the Details here')
st.write("""This Random Forest Regressor model helps to forecast unrolled values with impressive accuracy. 
            Leveraging the strength of the Random Forest technique, we can now make reliable predictions that 
            enable us to plan and strategize effectively in the fast-paced media landscape.""")

# load the saved model using pickle
with open('aajTak_model.pkl', 'rb') as file:
    model = pickle.load(file)

# # # load the saved model using joblib
# model3 = joblib.load('aajTak_model.joblib')

# Load the saved weekDay label encoder object using pickle
with open('weekDay_le.pkl','rb') as file1:
    weekDay_le = pickle.load(file1)

# Load the saved timeBand label encoder object using pickle
with open('timeBand_le.pkl','rb') as file2:
    timeBand_le = pickle.load(file2)

# previous_number_of_repairs = 
# st.sidebar.number_input('Enter the Previous Number of Repairs Undergone  0 to 5 )',min_value=0,max_value=5,step=1)  
    
# DATA from user
def user_report():
  
  # Share = round(float(st.sidebar.slider('Share', 0.000000, 100.000000, 0.611246, step=0.000001)), 6)
  # AMA = round(float(st.sidebar.slider('AMA', 0.000000, 45.000000, 4.196084, step=0.000001)), 6)
  # rate = round(float(st.sidebar.slider('rate', 0.000000, 1.500000, 0.018516, step=0.000001)), 6)  
  # daily_reach = round(float(st.sidebar.slider('daily reach', 0.000000, 300.000000, 36.23)), 6) 
  # cume_reach = round(float(st.sidebar.slider('cume reach', 0.000000, 300.000000, 36.231006)), 6)

    
  Share = round(float(st.sidebar.number_input('Share', 0.0, 100.0, 0.611246, step=0.000001)), 6)
  AMA = round(float(st.sidebar.number_input('AMA', 0.0, 45.0, 4.196084, step=0.000001)), 6)
  rate = round(float(st.sidebar.number_input('rate', 0.0, 1.5, 0.018516, step=0.000001)), 6)
  daily_reach = round(float(st.sidebar.number_input('daily reach', 0.0, 300.0, 36.23, step=0.000001)), 6)
  cume_reach = round(float(st.sidebar.number_input('cume reach', 0.0, 300.0, 36.231006, step=0.000001)), 6)

  

  # Share = st.sidebar.slider('Share', 0, 100, 0)
  # AMA = st.sidebar.slider('AMA', 0, 45, 4)
  # rate = st.sidebar.slider('rate', 0, 1, 0)
  # daily_reach = st.sidebar.slider('daily reach', 0, 300, 36)
  # cume_reach = st.sidebar.slider('cume reach', 0, 300, 36)
    
  # Output: {'Friday': 0, 'Monday': 1, 'Saturday': 2, 'Sunday': 3, 'Thursday': 4, 'Tuesday': 5, 'Wednesday': 6}
    
  Week_Day_Encoded = st.sidebar.selectbox("Week Day",
        ("Monday", "Tuesday","Wednesday","Thursday","Friday", "Saturday", "Sunday" ))
  if Week_Day_Encoded=='Monday':
      Week_Day_Encoded=1
  elif Week_Day_Encoded=="Tuesday":
      Week_Day_Encoded=5
  elif Week_Day_Encoded=="Wednesday":
      Week_Day_Encoded=6
  elif Week_Day_Encoded=="Thursday":
      Week_Day_Encoded =4
  elif Week_Day_Encoded=="Friday":
      Week_Day_Encoded =0
  elif Week_Day_Encoded=="Saturday":
      Week_Day_Encoded =2
  else:
      Week_Day_Encoded=3


  # The Time Band dictionary provided
  time_band_dict = {
    '02:00:00 - 02:30:00': 0, '02:30:00 - 03:00:00': 1, '03:00:00 - 03:30:00': 2, '03:30:00 - 04:00:00': 3,
    '04:00:00 - 04:30:00': 4, '04:30:00 - 05:00:00': 5, '05:00:00 - 05:30:00': 6, '05:30:00 - 06:00:00': 7,
    '06:00:00 - 06:30:00': 8, '06:30:00 - 07:00:00': 9, '07:00:00 - 07:30:00': 10, '07:30:00 - 08:00:00': 11,
    '08:00:00 - 08:30:00': 12, '08:30:00 - 09:00:00': 13, '09:00:00 - 09:30:00': 14, '09:30:00 - 10:00:00': 15,
    '10:00:00 - 10:30:00': 16, '10:30:00 - 11:00:00': 17, '11:00:00 - 11:30:00': 18, '11:30:00 - 12:00:00': 19,
    '12:00:00 - 12:30:00': 20, '12:30:00 - 13:00:00': 21, '13:00:00 - 13:30:00': 22, '13:30:00 - 14:00:00': 23,
    '14:00:00 - 14:30:00': 24, '14:30:00 - 15:00:00': 25, '15:00:00 - 15:30:00': 26, '15:30:00 - 16:00:00': 27,
    '16:00:00 - 16:30:00': 28, '16:30:00 - 17:00:00': 29, '17:00:00 - 17:30:00': 30, '17:30:00 - 18:00:00': 31,
    '18:00:00 - 18:30:00': 32, '18:30:00 - 19:00:00': 33, '19:00:00 - 19:30:00': 34, '19:30:00 - 20:00:00': 35,
    '20:00:00 - 20:30:00': 36, '20:30:00 - 21:00:00': 37, '21:00:00 - 21:30:00': 38, '21:30:00 - 22:00:00': 39,
    '22:00:00 - 22:30:00': 40, '22:30:00 - 23:00:00': 41, '23:00:00 - 23:30:00': 42, '23:30:00 - 24:00:00': 43,
    '24:00:00 - 24:30:00': 44, '24:30:00 - 25:00:00': 45, '25:00:00 - 25:30:00': 46, '25:30:00 - 26:00:00': 47}

  selected_time_band = st.sidebar.selectbox('Time Band', list(time_band_dict.keys()))
  Time_Band_Encoded = time_band_dict[selected_time_band]
    
  user_report_data = {
       'Share': Share, 
       'AMA': AMA, 
       'rate': rate, 
       'daily reach': daily_reach, 
       'cume reach': cume_reach, 
       'Week_Day_Encoded': Week_Day_Encoded, 
       'Time_Band_Encoded': Time_Band_Encoded}
  report_data = pd.DataFrame(user_report_data, index=[0])
  
  return report_data    
    
#Customer Data
user_data = user_report()
st.subheader("Entered Details")
st.write(user_data)


# define the prediction function
def predict_unrolled_value(user_data):
    
    # make the prediction using the loaded model and input data
    predicted_unrolled_value = model.predict(user_data)
    
    # # return the predict_unrolled_value as output
    # return predicted_unrolled_value[0]

    # return the predicted unrolled value as output with 6 decimal places
    return float(predicted_unrolled_value[0])

    
# Function calling
y_pred = predict_unrolled_value(user_data)

# CSS code for changing color of the button
st.markdown("""
    <style>
    .stButton button {
        background-color: #668f45;                        
        color: white;
    }
    </style>
    """, unsafe_allow_html=True)

# st.write("Click here to see the Predictions")
if st.button("Click here for Predictions"):
    st.subheader(f"Predicted Unrolled Value: {y_pred:.6f}")


# Testing purpose
# 0.611246	4.196084	0.018516	36.23	36.231006	'Saturday'	''08:00:00 - 08:30:00''
# 3.711884