Experimenting with artificial intelligence to predict house prices

Ho Chi Minh City A group of Vietnamese scientists collected nearly 200 housing data sets and put them into a machine learning model to predict housing prices close to the market.

This study was reported at the International Scientific Conference on Asian Urbanization (AUC2024) on the morning of January 12, organized by the Vietnam-Germany University. The representative of the research team, Dr. Tran Hong Ngoc, lecturer of computer science, faculty of engineering, Vietnam-Germany University, said that the demand for buying real estate for living and investment is currently very high. To refer to the current price, there are many information channels but it is difficult to determine the real value of the house. The research team wants to build a model to make the most realistic forecasts.

An apartment project in Ho Chi Minh City is selling in January 2024. Photo: Nguyen Tieu — An apartment project in Ho Chi Minh City is selling in January 2024. Photo: *Nguyen Tieu*

The experts used nearly 200 data sets of houses in Go Vap District, Ho Chi Minh City. The group used their knowledge and experience to process and clean the data. In some cases, the experts directly visited and assessed the value of the house to put into the database.

The cleaned dataset is divided into 23 attributes, showing house characteristics such as location, area, number of floors, number of rooms, year of construction, house type, interior and exterior design, repair status, surrounding amenities...

The data is grouped and fed into a machine learning model to predict house prices with 98% accuracy. "Currently, the data set is limited in size, so the team will soon collect more data on townhouses in Thu Duc City, Binh Duong ...", said Dr. Ngoc. The team believes that the value of a house will change when there are changes in surrounding amenities such as roads, bridges, etc., so it is necessary to continuously update data for the machine learning model so that the house price is always close to reality.

When users enter information describing the house with 23 attributes, the AI system will give a price prediction. However, Dr. Ngoc said that the research was only conducted for 6 months and it takes time to build a website as a platform to use the AI model. In addition, the group's AI model does not have data on planning, red books, etc. This is considered an important indicator to determine the legality and value of the house.

According to Dr. Ngoc, building an AI model to predict housing prices helps transparently manage sustainable and smart urban development. However, the machine learning model works accurately because the input data source must be correct and this depends largely on the impartiality of the manager.

The group’s research was among 40 presentations by more than 100 local and international scholars attending AUC. The topics of the presentations included: spatial transformation, smart cities, infrastructure and climate change, society and governance. The conference focused on discussing current issues and challenges in the field of urbanization research, while creating an opportunity for researchers, policy makers, and stakeholders to meet, share research and learn from experiences on diverse aspects of urbanization in Asia.

Dr. Thomas G. Aulig, Vice President of Vietnam-Germany University, said that the conference introduced many works of young scientists with research serving urban development. He hopes that AUC2024 will help the research community learn and share knowledge to strengthen academic networks and cooperation.

Ha An

Source link