摘要
对于加利福尼亚房价数据,房屋中位年龄可以看作是潜在的混淆变量,有可能影响其他协变量与响应变量之间的关系。如果忽略测量误差对变量的影响,并直接运用假设响应变量和协变量可以准确观测的经典半参数模型来拟合该数据,则可能会导致结果存在较大偏差,因此提出了利用单指标扭曲测量误差模型对该数据进行拟合。观察扭曲函数的拟合曲线后发现:中位房价、中位收入、总房间数、总卧室数和人口确实受到了以房屋中位年龄为混淆变量的乘积污染,这说明了选择的单指标扭曲测量误差模型相比于不考虑测量误差的半参数模型更适合加利福尼亚房价数据。
For California housing prices data,housing median age as a potential confounding variable may affect the relationship between other covariates and response variable.If we ignore the effect of measurement errors on the variables and directly apply the classic semi-parametric models assuming that the response variable and covariates can be accurately observed to fit the data,it may cause deviations in the results obtained.Therefore,we use single-index distortion measurement errors model to fit these data.After observing the fitted curve of the distortion function,we find that the confounding variable housing median age has a connection with the median house value,median income,total rooms,total bedrooms and population.This shows that the single-index distortion measurement errors model we choose is more suitable for California housing data than the semi-parametric models that do not consider measurement errors.
作者
娄文
LOU Wen(School of Science,Nanjing University of Science and Technology,Nanjing 210094,China)
出处
《重庆工商大学学报（自然科学版）》
2020年第6期95-102,共8页
Journal of Chongqing Technology and Business University:Natural Science Edition
关键词
单指标模型
乘积扭曲测量误差
加利福尼亚房价数据
single-index models
multiplicative distortion measurement errors
California housing prices data
