Rahma94/data_bender

1import streamlit as st2import pandas as pd3import numpy as np4import pickle5import matplotlib.pyplot as plt6from PIL import Image7from datetime import datetime, timedelta8from dateutil.relativedelta import relativedelta9 10# Load dataset11df = pd.read_csv('pro.csv')12 13# Convert the 1st column of the dataset into index14sales = df.set_index('Date')['Ship Quantity']15 16# Load model & scaler17with open('model.pkl', 'rb') as file1:18    model_lr = pickle.load(file1)19 20with open('scaler.pkl', 'rb') as file2:21    scaler = pickle.load(file2)22 23 24# Form for month input25 26def run():27 28    st.write('# Bend the Data')29    st.write('###### Enter the number of month and see how much product manufactured by then.')30 31    with st.form('forecastMonth'):32        month = st.number_input('Months (from December 2022)', min_value = 0, value = 4)33 34        # Submit button35        submitted = st.form_submit_button('Forecast')   36    37    # Function to predict38    def forecasting(month):39        sales_forecast = sales.copy()40        window = 441        for i in range(month):42            X = sales_forecast[-window:].values.reshape(1, -1)43            X_scaled = scaler.transform(X)44 45            last_month_str = sales_forecast.index[-1]46            last_month = datetime.strptime(last_month_str, '%y-%m')47 48            # Increment the month, handling overflow49            new_month = last_month + relativedelta(months=1)  # Increment by 1 month50 51            # Predict and round the sales value52            predicted_sales = round(model_lr.predict(X_scaled)[0])53 54            # Add a new row with the predicted sales and the new index55            new_row = pd.Series([predicted_sales], index=[new_month.strftime('%y-%m')], name='sales')56            sales_forecast = sales_forecast._append(new_row)57 58        # Create a DateTime index using the 'Date' column59        sales_forecast.index = pd.to_datetime(sales_forecast.index, format='%y-%m')60 61        # Drop the 'Date' column62        sales_forecast = sales_forecast.drop(columns=['Date'])63 64        return sales_forecast65    66    # If submit button is pressed67    if submitted:68        fig = plt.figure(figsize=(20,5))69        sales_forecast = forecasting(month)70        sales_forecast.index = sales_forecast.index.strftime('%y-%m')  # Convert Periods to strings71        sales_forecast.plot(color='blue', label='forecast', figsize=(20, 5))72 73        # Convert PeriodIndex to strings74        sales.index = sales.index.astype(str)75        sales.plot(color='red', label='real')76 77        plt.legend()78        plt.show()79        st.pyplot(fig)80 81        st.write(sales_forecast)82 83if __name__ == '__main__':84    run()85