train_data.csv

This dataset focuses on predicting long-term health behaviors in older adults. The dependent variable, "TTM Stage," represents the respondent's current behavioral stage based on the Transtheoretical Model (TTM), measured on a 5-point scale where 1 = precontemplation and 5 = maintenance. Independent variables cover multiple dimensions: In individual characteristics, "A1(Age)" records age as an open-ended numeric value; "A2(Gender)" uses 0 for male and 1 for female; "A3(Chronic Illness Status)" uses 0 to indicate no chronic disease and 1 for yes. For health app usage, "B1(Health App Type)" is categorized into 4 types (1 = voice assistant, 2 = virtual assistant, 3 = wearable device, 4 = others), and "B2(Weekly Usage Frequency)" is measured on a 4-point scale (1 = <1 time/week, 2 = 1–2 times/week, 3 = 3–4 times/week, 4 = daily). Based on the Health Belief Model (HBM), twelve 7-point Likert scale variables assess perceptions of susceptibility (C1-C2), severity (C3-C4), benefits (C5-C6), barriers (C7-C8), cues to action (C9-C10), and self-efficacy (C11-C12), reflecting respondents’ cognitions and attitudes toward health and health apps. Together, these variables provide comprehensive data support for random forest model analysis.

本数据集旨在预测老年人的长期健康行为。因变量“TTM阶段”基于跨理论模型（Transtheoretical Model，TTM）反映受访者当前的行为阶段，采用5点量表测量，其中1代表前沉思阶段，5代表维持阶段。自变量涵盖多个维度：在个体特征方面，“A1（年龄）”以开放式数值记录年龄；“A2（性别）”中0代表男性、1代表女性；“A3（慢性病状态）”中0表示无慢性病、1表示有慢性病。在健康应用使用方面，“B1（健康应用类型）”分为4类（1=语音助手，2=虚拟助手，3=可穿戴设备，4=其他）；“B2（每周使用频率）”采用4点量表测量（1=每周少于1次，2=每周1-2次，3=每周3-4次，4=每日）。基于健康信念模型（Health Belief Model，HBM），12个7点李克特量表变量评估了易感性（C1-C2）、严重性（C3-C4）、益处（C5-C6）、障碍（C7-C8）、行动线索（C9-C10）及自我效能感（C11-C12）的认知，反映受访者对健康及健康应用的认知与态度。这些变量共同为随机森林模型分析提供了全面的数据支持。