app.py

import pandas as pd
import pickle
import joblib
from datetime import datetime




def add_new_columns(df2):
    df = df2.copy()
    # easy to implement
    df['day_of_week'] = df['Date'].dt.dayofweek
    df['day_of_month'] = df['Date'].dt.day
    df['last_weekend'] = ((df['day_of_week'] == 4) | (df['day_of_week'] == 5) | (df['day_of_week'] == 6)) & (df['day_of_month'] + 7 > df['Date'].dt.days_in_month)
    df['Day_of_week_numeric'] = df['Date'].dt.dayofweek
    df['Day_of_year_numeric'] = df['Date'].dt.dayofyear
    df['Week_numeric'] = df['Date'].dt.isocalendar().week
    # df['Avg_Sale_Weekday'] = df.groupby(['Location', df['Date'].dt.dayofweek])['Number_of_burgers'].transform('mean')
    return df

def predict_without_columns(X, yhat, day,month,year,temp,rain,wind,sun,snow, day_of_the_week,location,parken_event, general_event,is_holiday,rf,gb,hgb,xgr,lgbr,reg):
    new_row = {name:0 for name in X}
    new_row['yhat'] = yhat
    new_row['Day'] = day
    new_row['Month'] = month
    new_row['Year'] = year
    new_row['Temperature'] = temp
    new_row['Wind'] = wind
    new_row['Rain'] = rain
    new_row['Sun'] = sun
    new_row['Snow'] = snow
    new_row['holiday'] = is_holiday
    new_row['Day_of_the_week_' + day_of_the_week] = 1
    new_row['Location_' + location] = 1
    new_row['Parken_Event_' + parken_event] = 1
    new_row['General_Event_' + general_event] = 1
    new_pd = pd.DataFrame([new_row])
    
    new_pd['Date'] = pd.to_datetime(new_pd[['Year', 'Month', 'Day']])
    new_pd = add_new_columns(new_pd)
    new_pd.drop('Date', axis=1, inplace=True)

    columns = ['yhat', 'Day', 'Month', 'Year', 'Temperature', 'Rain', 'Wind', 'Sun',
       'Snow', 'day_of_week', 'day_of_month',
       'last_weekend', 'Day_of_week_numeric', 'Day_of_year_numeric',
       'Week_numeric', 'holiday', 'Day_of_the_week_Friday',
       'Day_of_the_week_Monday', 'Day_of_the_week_Saturday',
       'Day_of_the_week_Sunday', 'Day_of_the_week_Thursday',
       'Day_of_the_week_Tuesday', 'Day_of_the_week_Wednesday', 'Location_DGH',
       'Location_HDM', 'Location_HPNS', 'Location_LGV', 'Location_NHG',
       'Location_VDV', 'Parken_Event_Concert', 'Parken_Event_Football',
       'Parken_Event_None', 'General_Event_Cultural', 'General_Event_Fashion',
       'General_Event_Festivitie', 'General_Event_None',
       'General_Event_Social', 'General_Event_Sport',
       'General_Event_Worldwide'] 
    new_pd = new_pd[columns]

    prediction_reg = reg.predict(new_pd)
    prediction_rf = rf.predict(new_pd)
    prediction_xgr = xgr.predict(new_pd)
    prediction_gb = gb.predict(new_pd)
    prediction_hgb = hgb.predict(new_pd)
    prediction_lgbr = lgbr.predict(new_pd)
    mean_pred = (prediction_reg  + prediction_xgr  + prediction_lgbr + prediction_hgb + prediction_gb) / 6
    output = 'On ' + str(day_of_the_week) + ' , ' + str(day) + ' - ' + str(month) + ' - ' + str(year) + ', the location ' + str(location) + ' is predicting to sell ' + str(round(mean_pred[0])) + ' burgers'
    return output

def predict(day,month,year,location,temp,rain,wind,sun,parken_event,general_event):
    snow=0
    day, month, year = int(day), int(month), int(year)
    temp, rain, wind, sun, snow = float(temp), float(rain), float(wind), float(sun), float(snow)
    location, parken_event, general_event = str(location), str(parken_event), str(general_event)


    date_obj = datetime(year, month, day)
    day_of_week_num = date_obj.weekday()
    days = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
    day_of_the_week = days[day_of_week_num]
    
    prophet_file = f'prophet_df_{location}.csv'
    df_prophet = pd.read_csv(prophet_file)
    df_prophet = df_prophet[['ds','yhat']].copy()
    df_prophet.set_index('ds',inplace=True)
    date_string = f'{year}-{month}-{day}'
    datetime_obj = str(datetime.strptime(date_string, '%Y-%m-%d').date())
    yhat = df_prophet.loc[datetime_obj]['yhat']

    holidays_dk = pd.read_csv('holidays_denmark.csv')
    holidays_dk = holidays_dk[['ds','holiday']].copy()
    holidays_dk.set_index('ds',inplace=True)
    is_holiday = 1 if datetime_obj in holidays_dk.index else 0
    print(is_holiday)
    
    rf = joblib.load("filename_rf.joblib")
    gb = joblib.load("filename_gb.joblib")
    hgb = joblib.load("filename_hgb.joblib")
    xgr = joblib.load("filename_xgr.joblib")
    lgbr = joblib.load("filename_lgbr.joblib")
    reg = joblib.load("filename_reg.joblib")

    X = ['yhat', 'Day', 'Month', 'Year', 'Temperature', 'Rain', 'Wind', 'Sun', 'Snow',
       'Day_of_the_week_Friday', 'Day_of_the_week_Monday',
       'Day_of_the_week_Saturday', 'Day_of_the_week_Sunday',
       'Day_of_the_week_Thursday', 'Day_of_the_week_Tuesday',
       'Day_of_the_week_Wednesday', 'Location_DGH', 'Location_HDM',
       'Location_HPNS', 'Location_LGV', 'Location_NHG', 'Location_VDV',
       'Parken_Event_Concert', 'Parken_Event_Football', 'Parken_Event_None',
       'General_Event_Cultural', 'General_Event_Fashion',
       'General_Event_Festivitie', 'General_Event_None',
       'General_Event_Social', 'General_Event_Sport',
       'General_Event_Worldwide']
    actual_prediction = predict_without_columns(X,yhat,day,month,year,temp,rain,wind,sun,snow,day_of_the_week,location,parken_event, general_event,is_holiday,rf,gb,hgb,xgr,lgbr,reg)
    return actual_prediction







import gradio as gr

theme = gr.themes.Soft(
    primary_hue="green",
    secondary_hue="emerald"
)

# Launch a Gradio interface
title = 'Predicting number of burgers - Gasoline Grill'
description = 'This is an AI model that predicts the number of burgers based on different parameters. These parameters are:\n1. Day: current day of the month (from 1 to 31)\n2. Month: Current month (from 1 to 12)\n3. Year: current year.\n4. Day of the week: current day of the week (from Monday to Sunday)\n5. Location: current location (LGV, NHG, HDM, VDV, DGH, HPNS)\n6. Temperature: expected medium temperature of the day (in ºC)\n7. Rain: expected rain of the day (in mm)\n8. Wind: expected wind of the day (in m/s)\n9. Sun: expected hours of sunlight of the day (in hours)\n10. Parken event: Is there an event in Parken? (Football, Concert, None)\n11. General Event: Is there a general event in Copenhagen? (Cultural, Fashion, Festivity, None, Social, Sport, Worldwide)\n\nFor temperature, wind, rain and sun, check the DMI webpage https://www.dmi.dk/vejrarkiv or check the weather app on your phone :)\n\nOnce you have all the data, you must put it in order, separated by commas and without spaces between the data.\n\nSome examples are shown below. Click on them to see a real example'
examples = ['12,2,2024,Monday,VDV,1.8,5.6,2.4,0,None,None','29,5,2024,Wednesday,LGV,10,2,1.4,5,Football,Social']

# demo = gr.Interface(fn=predict, title=title, description=description)

with gr.Blocks(theme=theme) as demo:
    with gr.Row():
        gr.Markdown(
        """
        # Predicting number of burgers - Gasoline Grill 🍔
        This is an AI model that predicts the number of burgers to be sold based on different parameters. These parameters are:
        1. Day, Month , Year
        2. Location (LGV, DGH, NHG, HDM, VDV, HPNS)
        3. Temperature: expected medium temperature of the day (in ºC)
        4. Rain: expected rain of the day (in mm)
        5. Wind: expected wind of the day (in m/s)
        6. Sun: expected hours of sunlight of the day (in hours)
        7. Parken event: Is there an event in Parken?
        8. General Event: Is there a general event in Copenhagen? 
        
        For temperature, wind, rain and sun, check the DMI webpage https://www.dmi.dk/vejrarkiv or the weather app on your phone :)
        
        Once you have all the data, click the Predict button.    
        """
        )
    with gr.Row():
        day = gr.Dropdown([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,21,22,23,24,25,26,27,28,29,30,31],label='Day')
        month = gr.Dropdown([1,2,3,4,5,6,7,8,9,10,11,12],label='Month')
        year = gr.Dropdown([2024],label='Year', value=2024)
        # day_of_the_week = gr.Dropdown(['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'], label='Day of the week')
        location = gr.Dropdown(['LGV', 'DGH','NHG' ,'HDM', 'VDV','HPNS'], label='Location')
    with gr.Row():
        with gr.Column():
            temp = gr.Slider(minimum=-10, maximum=40, label='Temperature (°C)')
            sun = gr.Slider(minimum=0, maximum=17, label='Sunlight (hours)')
        with gr.Column():
            rain = gr.Slider(minimum=0, maximum=30, label='Rain (mm)')
            wind = gr.Slider(minimum=0, maximum=12, label='Wind (m/s)')
            # snow = gr.Slider(minimum=0, maximum=3, label='Snow')
    
    with gr.Row():
        parken_event = gr.Dropdown(['Football', 'Concert', 'None'], label='Parken Event', value='None')
        general_event = gr.Dropdown(['Cultural', 'Fashion', 'Festivity', 'None', 'Social', 'Sport', 'Worldwide'], label='General Event', value='None')
    
    with gr.Row():
        predict_2 = gr.Button(value = 'Predict')
    
    with gr.Column():
        prediction = gr.Textbox()
    
    predict_2.click(predict, inputs = [day,month,year,location,temp,rain,wind,sun,parken_event,general_event], outputs = prediction)
    
# demo.launch(auth=('gasolgrill','gasolgrill'))
demo.launch()


# gr.Interface(predict, 
#              inputs = [
#                  gr.Row([gr.Textbox(label='Day'),gr.Textbox(label='Month'),gr.Textbox(label='Year'),gr.Dropdown(['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'], label='Day of the week'),gr.Dropdown(['LGV', 'NHG', 'DGH', 'HDM', 'VDV'], label='Location')]),
#                  gr.Row([gr.Slider(minimum=-10, maximum=40, label='Temperature (°C)'), gr.Slider(minimum=0, maximum=17, label='Sunlight (hours)'), gr.Slider(minimum=0, maximum=30, label='Rain (mm)'), gr.Slider(minimum=0, maximum=12, label='Wind (m/s)')]),
#                  gr.Row([gr.Dropdown(['Football', 'Concert', 'None'], label='Parken Event'), gr.Dropdown(['Cultural', 'Fashion', 'Festivity', 'None', 'Social', 'Sport', 'Worldwide'], label='General Event')]),
#                  ], 
#              outputs="textbox", 
#              title=title, description = description
#             ).launch(share=True)