Data pertaining to Chapter 5 "A Novel Framework for Understanding and Identifying Driving Heterogeneity through Action Patterns"

This dataset supports the paper <em>“A Novel Framework for Understanding and Identifying Driving Heterogeneity through Action Patterns”</em> (Chapter 5 of the PhD dissertation). The study focuses on data analysis and behavioural modelling, introducing a new framework for identifying driving heterogeneity based on underlying action patterns in driver behaviour. The framework includes three processes: Action phase extraction, Action pattern calibration, and Action pattern classification. Evaluation of the framework on a large-scale naturalistic driving dataset reveals six distinct Action patterns. The implementation of the attention mechanism to LSTM models significantly enhanced both the accuracy and time efficiency of Action pattern identification. The data was generated and processed using rule-based segmentation, unsupervised learning, feature extraction, and supervised learning techniques in Python. It is provided as a zipped folder containing two subfolders, with files in <code>.xlsx</code>, <code>.csv</code>, <code>.mat</code>, <code>.m</code>, <code>.txt</code>, and <code>.pdf</code> formats. A <code>ch5_Readme.txt</code> file is included to explain the structure of the data and provide instructions for use.

本数据集支撑论文《基于动作模式的驾驶异质性理解与识别新框架》（该论文为博士学位论文第5章）。本研究聚焦数据分析与行为建模领域，提出了一种基于驾驶员行为底层动作模式的驾驶异质性识别新框架。该框架包含三大流程：动作阶段提取、动作模式校准与动作模式分类。在大规模自然驾驶数据集上对该框架开展评估后，共得到六种差异化的动作模式。将注意力机制应用于长短期记忆网络（LSTM）模型后，显著提升了动作模式识别的准确率与时间效率。本数据集通过Python中的基于规则的分割、无监督学习、特征提取与监督学习技术生成并处理。该数据集以压缩包形式提供，内含两个子文件夹，文件格式涵盖.xlsx、.csv、.mat、.m、.txt与.pdf。压缩包中包含ch5_Readme.txt文件，用于说明数据集结构并提供使用说明。