Airbnb Price Prediction and Revenue Forecasting Using Machine Learning

Project Title

Airbnb Price Prediction and Revenue Forecasting Using Machine Learning

Project Area of Specialization Artificial IntelligenceProject Summary

Airbnb is an online marketplace for short-term home and apartment rentals. It offers a service for people to rent out their homes or living space for a short period while they are away or spare space to travelers. While Airbnb is an exciting service to earn some extra cash while renting out their home, it is very challenging for property owners to determine the price of the property as Airbnb determines the price based on the number of guests.

This work aims to develop a model for machine learning, which will predict the price of the property using given information and forecast revenue based on other rental information in the locality. Airbnb historical data will be scrapped and used to train two types of Machine Learning models 1) Hedonic model Regression, and 2) XGBoost and compare their results for best accuracy. These Machine Learning models will later be hosted on a web server. A mobile compatible web client will be able to send queries to the webserver with the address of the property for 1) Property Rent price prediction, 2) Periodic revenue forecast, and 3) features of super-host in the neighborhood. The final product of this project will be a web application that Airbnb property renters will able to use for the features as mentioned earlier.

Project Objectives

1. Scrap data for training Machine Learning model
2. Train Machine Learning model for price prediction
3. Train Machine Learning model for revenue forecasting
4. Train Machine Learning model to find features of Superhost
5. Develop a web server for hosting Machine Learning model
6. Develop Rest APIs that will be used from client application to send and retrieve data from an application server
7. Develop a front-end web application for following result visualization
   • Price Prediction for the user input address
   • Revenue forecast for the user input address
   • Features of Superhost in the user’s neighborhood

Project Implementation Method

This project divides into two parts:

1. Model Development
2. Web Application

Model Development

In this work, I will develop a machine learning model to predict the price of the property. For training the machine learning model.

1.    Data Collection
I will use Airbnb data, which can be accessed at this portal: Inside Airbnb. This portal provides scrapped data for many popular cities around the world. Additional data can be scrapped using Airbnb Data Collection, an open-source tool available of GitHub.

2.    Designing Model
I will train two different types of Machine learning models and compare their accuracy and performance:

• Hedonic Model Regression: It is a common technique used in real estate appraisal and real estate economics. This model will produce hedonic price estimation that includes spatial and locational features.
• XGBoost model: This is an implementation of gradient boosted decision trees.

3.    Implementation
I will develop these models using Python programming language on Jupiter Notebook. For evaluation metrics, I will use mean squared error (for loss) and r-squared (for accuracy).

Web Application

A client-server web application will develop to get user input, test machine learning models, and show results.

1.    Server Application
I will develop a server application where the Machine Learning model will host as well. This server application will provide REST APIs for getting information from the client application and result for visualization.

2.    Client Application
A web client application will develop, which can take input from a user, send queries to web servers, get results, and visualize them. This web client application will be mobile-friendly and will be made with focus to be useable on mobile phones.
 

Benefits of the Project

The final product of this project will be a web application where the Machine Learning model will be hosting to make property rent prediction and revenue forecasting. Airbnb renters will be able to use this application for getting price prediction and revenue forecasts for their property. Also, this application will highlight features of super-host in the locality of a user, learning from this information they will be able to improve host-experience and thus value for their property.

Technical Details of Final Deliverable Machine Learning Model

The machine learning model will be developed using the Python programming language. This model will use PyTorch as a Machine Learning library. For model performance and accuracy graphs, I will use Matplotlib library.

Web Application

A client-server application development will also be part of this project.

The server-side application will be developed in the Python programming language. This server-side application will have access to the database and the Machine Learning model from previous step. This application will project an API interface for querying requests from the web-client application.

The web-client application will use the following technologies:

• HTML5
• Bootstrap
• CSS
• SASS
• JavaScript
• D3JS
• AngularJS

This web-client will use an API interface to send requests to the server-side application to get price prediction, revenue forecast, features of super-host in the locality and other user-related operations and present the data with the help of Javascript libraries.

Final Deliverable of the Project Software SystemCore Industry OthersOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies Big DataSustainable Development Goals Decent Work and Economic Growth, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	60000
Google Collab for training Machine Learning models	Equipment	1	25000	25000
Stationary	Miscellaneous	1	5000	5000
Printing	Miscellaneous	1	5000	5000
Web Hosting with database	Equipment	1	25000	25000