Midterm Project - MLZoomcamp

Dataset Kaggle Community prediction Competition

hotel booking cancellations

Source

Hotel booking demand datasets

Overview

The 79 DAPT SAO IH Hotel Booking describes two datasets with hotel demand data. One of the hotels is a resort hotel and the other is a city hotel. Both datasets share the same structure. Each observation represents a hotel booking. Both datasets comprehend bookings due to arrive between the 1st of July of 2015 and the 31st of August 2017, including bookings that effectively arrived and bookings that were canceled.

Attribute Information

Name	Data Type	Description
hotel	category	Type of hotel
is_cancelled	binary	Target Value indicating if the booking was canceled (1) or not (0)
lead_time	number	Number of days that elapsed between the entering date of the booking into the PMS and the arrival date
stays_in_weekend_nights	number	Number of weekend nights (Saturday or Sunday) the guest stayed or booked to stay at the hotel
stays_in_week_nights	number	Number of week nights (Monday to Friday) the guest stayed or booked to stay at the hotel
adults	number	Number of adults
children	number	Number of children
babies	number	Number of babies
meal	category	Type of meal booked
country	category	Country of origin
market_segment	category	Market segment designation
distribution_channel	category	Booking distribution channel
is_repeated_guest	binary	Value indicating if the booking name was from a repeated guest (1) or not (0)
previous_cancellations	number	Number of previous bookings that were cancelled by the customer prior to the current booking
previous_bookings_not_canceled	number	[Number of previous bookings not cancelled by the customer prior to the current booking
reserved_room_type	category	Code of room type reserved
assigned_room_type	category	Code for the type of room assigned to the booking
booking_changes	number	Number of changes/amendments made to the booking
deposit_type	category	Indication on if the customer made a deposit to guarantee the booking
agent	category	ID of the travel agency that made the booking
company	category	ID of the company/entity that made the booking or responsible for paying the booking. ID is presented instead of designation for anonymity reasons
days_in_waiting_list	number	Number of days the booking was in the waiting list before it was confirmed to the customer
customer_type	category	Type of booking
adr	number	Average Daily Rate calculated by dividing the sum of all lodging transactions by the total number of staying nights
required_car_parking_spaces	number	[Number of car parking spaces required by the customer
total_of_special_requests	number	Number of special requests made by the customer
reservation_status_date	date	Date at which the last status was set
arrival_date	date	Date of arrival
id_booking	number	ID of booking

Problem Statement

Booking cancellations often directly impact a hotel's bottom line: many times the room(s) reserved are not booked again, resulting in lower occupation and, subsequently, revenue. To hedge this risk, hotels often demand a booking deposit, often calculated as a percentage of the reservations full price. However, this practice can directly impact demand as some customers might look for different hotels with no deposit (or a smaller one). Our goal is to build a model that, by predicting whether a booking will be cancelled or not, can be used by the hotel to implement different risk-aware strategies for calculating the deposit size.

Contents of the folder

The input folder contains only the cleaned dataset. For original datset please refer to the notebook.

The notebooks folder contains the notebook.ipynb Which contains:

• EDA
• Feature enginnering
• Comparison of 8 models, namely:
  • the basic model Dummy
  • Decision Tree ,
  • Gaussian Naive Bayes,
  • Random Forest,
  • Historgram Gradient Boosting,
  • XGBoost,
  • Logistics Regression,
  • KNeighbors
• Model Optimization
• Fitting the best Model(s)
• Evaluation of the best Model(s)
• Feature Importance and extract the high features.
• ROC Curve
• Evaluation of Model(s) with high features

The model is fitted with clean data and saved in binary mode in train.py file.

Deployment of model

I am using Flask/gunicorn on linux ubuntu, in order to deploy the model. To deploy this model with Flask/gunicorn, please use:

pipenv run gunicorn --bind 0.0.0.0:9696 predict:app

Virtual Environment/venv

I used pipenv for the virtual environment. In order to use the same venv as me, do use:

pip install pipenv

To replicate the environment, on your command line, use

pipenv install numpy pandas scikit-learn flask gunicorn requests

Note:
If you want to train the xgboost model instead of Histogram Gradient Boosting please add xgboost as command on pipenv installation.

Docker

Note:
To perform the following steps you should logon to your DockerHub Account ( Logi & Password)

I have built the model and pushed it to dajebbar/hotel-booking-model:latest. To use it just

docker pull dajebbar/hotel-booking-model:latest

Or in order to take the model from the docker container I built, just replace

FROM python:3.9-slim 

#with 

FROM dajebbar/hotel-booking-model:latest in the dockerfile.

If you choose to build a docker file locally instead, here are the steps to do so:

1. Create a Dockerfile as such:

FROM python:3.9-slim

ENV PYTHONUNBUFFERED=TRUE

RUN pip --no-cache-dir install pipenv

WORKDIR /app

COPY ["Pipfile", "Pipfile.lock", "./"]

RUN pipenv install --deploy --system && rm -rf /root/.cache

COPY ["predict.py", "*.bin", "./"]

EXPOSE $PORT

CMD gunicorn --workers=4 --bind 0.0.0.0:$PORT predict:app

This allows us to install python, run pipenv and its dependencies, run our predict script and our model itself and deploys our model using Flask/gunicorn.

Similarly, you can just use the dockerfile in this repository.

2. Build the Docker Container with :

 docker build -t hotel-booking-model .

3. Run the Docker container with:

Docker run -it -p 9696:9696 docker build -t hotel-booking-model

Now we can use our model through

python test_pred.py

4. tag the docker container with:

docker tag hotel-booking-model dajebbar/hotel-booking-model:latest

5. Push it Docker registry with :

docker push dajebbar/hotel-booking-model:latest

Heroku

In order to deploy this model on heroku cloud server, here are the steps to follow:

• Create a new .github folder in the parent folder, then in the latter, create another workflows folder, you can do it easily using the linux command:

touch .github/workflows

• in the workflows folder, create the main.yaml file, which is responsible for auto-reading the Dockerfile file once you push it to github. Here is its content:

# Your workflow name.
name: Deploy to heroku.

# Run workflow on every push to main branch.
on:
  push:
    branches: [main]

# Your workflows jobs.
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      # Check-out your repository.
      - name: Checkout
        uses: actions/checkout@v2


### ⬇ IMPORTANT PART ⬇ ###

      - name: Build, Push and Release a Docker container to Heroku. # Your custom step name
        uses: gonuit/heroku-docker-deploy@v1.3.3 # GitHub action name (leave it as it is).
        with:
          # Below you must provide variables for your Heroku app.

          # The email address associated with your Heroku account.
          # If you don't want to use repository secrets (which is recommended) you can do:
          # email: my.email@example.com
          email: ${{ secrets.HEROKU_EMAIL }}
          
          # Heroku API key associated with provided user's email.
          # Api Key is available under your Heroku account settings.
          heroku_api_key: ${{ secrets.HEROKU_API_KEY }}
          
          # Name of the heroku application to which the build is to be sent.
          heroku_app_name: ${{ secrets.HEROKU_APP_NAME }}

          # (Optional, default: "./")
          # Dockerfile directory.
          # For example, if you have a Dockerfile in the root of your project, leave it as follows:
          dockerfile_directory: ./

          # (Optional, default: "Dockerfile")
          # Dockerfile name.
          dockerfile_name: Dockerfile

          # (Optional, default: "")
          # Additional options of docker build command.
          docker_options: "--no-cache"

          # (Optional, default: "web")
          # Select the process type for which you want the docker container to be uploaded.
          # By default, this argument is set to "web".
          # For more information look at https://devcenter.heroku.com/articles/process-model
          process_type: web

• Then, go to github, open the folder hotel_booking_project then click on Settings, go to Secrets then click on Actions, then click on the button New repository secret. On the Name field type HEROKU_EMAIL , below there is the Secret field, enter your email which serves as your identifier on Heroku. After confirming by clicking on the Add secret button, you must repeat the same manipulation this time type HEROKU_API_KEY as name, and go to the Heroku platform, go to Accounts settings then go down to the API Key field then click on the Reveal button copy-paste your API Key into the Secret field in github. The same manipulation this time on the Name field type HEROKU_APP_NAME and in the Secret field type the name you gave to your application on Heroku, in my case it is called hotelbooking-api. Once saved, go back to your .github folder, open the console, type:

git add .
git commit -m "your comment"
git push main origin

And now, when you refresh your github page, a small yellow dot will appear next to the folder name, click on commits to see the automatic execution of the Dockerfile file with all its dependencies. Return to the Heroku platform, press the Open app button and voila.

Test the project

The project is now deployed on heroku cloud servers and to test it just run the file heroku_pred.py.

Want to Contribute?

• Fork 🍴 the repository and send PRs.
• Do ⭐ this repository if you like the content.

Connect with me: