基于Logistic回归和平均贝叶斯网络的人员开窗行为研究*
Research on personnel window opening behavior based on Logistic regression and average Bayesian network models
摘要：
对天津某高校宿舍内的人员开窗行为进行了一个完整供暖季的监测。对传统Logistic开窗预测模型的输入参数进行了简化，提出了预测准确度较高且更具实用价值的简化Logistic回归模型。并将平均贝叶斯网络模型引入开窗行为的预测中，取得了较好的预测效果，模型预测准确率为82.22%，其中开窗的预测准确率比Logistic回归模型提高14.16%，体现了平均贝叶斯网络模型在开窗行为预测中的优越性。
Abstract：
Investigates the personnel window opening behaviors in the dormitories of a university in Tianjin during a whole heating season. Simplifies the input parameters of the traditional Logistic window prediction model, and presents a simplified Logistic regression model with higher prediction accuracy and practicability. Predicts the window opening behavior by the average Bayesian network model, and obtains a better prediction effect, with 82.22% of the prediction accuracy. The prediction accuracy of window opening of the average Bayesian network model is 14.16% higher than that of the Logistic regression model, which reflects the superiority of the average Bayesian network model in the prediction of window opening behaviors.
Keywords:average Bayesian network, Logistic regression, prediction model, window opening behavior, window opening probability