Scholarlink

This paper proposes a novel method for clustering multivariate time series data using a combination of dimensionality reduction and entropy-aided clustering. The method effectively captures patterns and trends in the data, allowing for accurate clustering of similar time series segments in various fields where multivariate time series data is prevalent. Additionally, this paper aims to improve the accuracy of real estate price prediction during uncertain events by proposing an effective framework, including three main components: sentiment analysis using natural language processing, clustering using the Jensen-Shannon Entropy-aided Distance Clustering (JSDC) method, and forecasting using a Sentiment Analysis Enhanced clustering Forecasting (SECF) neural network. Notably, the integration of sentiment analysis within the JSDC method contributes significantly to the framework’s capability to capture and incorporate market sentiment, thereby enhancing clustering accuracy and forecasting performance. Extensive experiments on monthly real estate prices and headline news datasets spanning over 20 years demonstrate that our SECF framework achieves notable improvements in accuracy. Specifically, compared to traditional methods like ARIMA, the SECF framework shows reductions in MAPE (from 3708.4 to 11.5), RMSE (from 58.9 to 10.7), and MAE (from 58.7 to 10.0), highlighting its superior forecasting performance and robustness in handling complex multivariate time series data.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Australian Real Estate Market Price Prediction with Entropy Clustering and Sentiment-Based Deep Neural Framework

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dat Le,
  • Sutharshan Rajasegarar,
  • Wei Luo,
  • Thanh Thi Nguyen,
  • Maia Angelova

摘要

This paper proposes a novel method for clustering multivariate time series data using a combination of dimensionality reduction and entropy-aided clustering. The method effectively captures patterns and trends in the data, allowing for accurate clustering of similar time series segments in various fields where multivariate time series data is prevalent. Additionally, this paper aims to improve the accuracy of real estate price prediction during uncertain events by proposing an effective framework, including three main components: sentiment analysis using natural language processing, clustering using the Jensen-Shannon Entropy-aided Distance Clustering (JSDC) method, and forecasting using a Sentiment Analysis Enhanced clustering Forecasting (SECF) neural network. Notably, the integration of sentiment analysis within the JSDC method contributes significantly to the framework’s capability to capture and incorporate market sentiment, thereby enhancing clustering accuracy and forecasting performance. Extensive experiments on monthly real estate prices and headline news datasets spanning over 20 years demonstrate that our SECF framework achieves notable improvements in accuracy. Specifically, compared to traditional methods like ARIMA, the SECF framework shows reductions in MAPE (from 3708.4 to 11.5), RMSE (from 58.9 to 10.7), and MAE (from 58.7 to 10.0), highlighting its superior forecasting performance and robustness in handling complex multivariate time series data.