| 摘要： | Research on the study of houses, condominiums and buildings in Taiwan’s metropolitan areas
continues to be an important area of research. In real estate forecasting and analysis, methods such as
statistical analysis and questionnaire data collection are widely used. However, when multidimensional data
is considered, these methods are time-consuming and inadequate. This study aimed to build a real estate
forecasting model that can adapt to a changing environment. Data were collected from public government
databases, the collected data were standardized for accurate clustering, an appropriate data clustering
algorithm was applied to the standardized data, and cross-statistical analysis was performed to verify the
adopted algorithm. We used a deep learning based on autoencoder algorithm to increase the accuracy of the
clustering analysis. A double-bottom map particle swarm optimization (DBM-PSO) clustering algorithm was
then used to determine the optimal clustering solution. Cluster analysis and deep learning were conducted
on data collected from public websites to understand the factors that led to the sustained increase in housing
prices in Taiwan over the past decade. The results of this study indicate that three key factors—the number
of real estate transactions, the average unit price of real estate transactions, and the building material and
construction index—significantly affected real estate prices in Taiwan. Our results could help researchers
and governments to focus on specific aspects of real estate development without being influenced by other
related factors. In addition, the relationships between real estate trends and the aforementioned three key
factors were determined to obtain valuable information that can enable the Taiwanese government to regulate
the property market and prevent excessive growth. The framework proposed in this paper allows researchers
and governments to focus on specific aspects of real estate development without being influenced by other
related factors, and provides a new mechanism for approaching real estate-related forecasting. |
