Housing price prediction using neural networks

• Main Author: Chan, Fung Foong
• Other Authors: Wang Lipo
• Format: Final Year Project
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/61073
• Institution: Nanyang Technological University

This research applies the artificial neural network (ANN) models to predict the public housing prices in Singapore. The study consists of two major sections. In Section 1, static ANN is used to estimate the selling price based on the housing characteristics; In Section 2, the dynamic ANN is used to estimate the trend of resale price index (RPI), with nine independent economic and demographic variables. Quarterly time series data from 1990 to 2013 are used for the ANN training, validation and testing. The results show that the ANN model is able to produce a good fit and predictions, as the Regression values (R-value) are higher than 0.9 in most cases. However, there are also significant problems when using the ANN models, such as the inability to conclude for optimum results due to the fluctuation of the predicted values. It is suggested that ANN is a suitable tool for forecasting property prices because of its capability to map the non-linear relationship between variables. Nonetheless, users should also be cautious of the potential issues, when using ANN models for any financial market predictions.