Application of Transformer Model in Predicting Housing Price in Selected Vietnamese Provinces

Abstract

This paper selects an appropriate machine learning model to forecast real estate and apartment prices across several major provinces and cities in Vietnam. While we primarily deploy the Transformer, we also experimented with other machine learning algorithms, such as Linear Regression, Decision Tree, and Random Forest, to demonstrate the Transformer's superior performance. We automatically collected raw data from Vietnam's largest real estate website and then preprocessed, analysed, and evaluated it. The analysis revealed that building a prediction model on a single dataset yields fewer effective results than splitting the data into two distinct sets, one for landed properties and the other for apartments, and developing separate models for each product category. Building two independent prediction models allows effective use of each product type's unique characteristics, improving the accuracy and efficiency of machine learning for real estate price prediction.