Regression calibration estimation of measurement errors in negative binomial regression models

• Main Authors: Huan-Su Wu, 吳桓夙
• Other Authors: 黃文瀚
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/91515714398562092507

碩士 === 國立中興大學 === 統計學研究所 === 105 === For the count data, often using the Poisson Regression or Negative Binomial Regression Model fit in medicine, public health and ecological. In this paper, we discuss the estimation method when the independent variable is a time series data and has the measurement error in the Negative Binomial Regression Model. We propose to use the Nonparametric Regression Model estimating the trend of the time series data in place of the observed data, and then further use the Regression Calibration to correct the estimation of the parameters. Therefore, the simulation study we compare the merits of the three methods in difference parameter settings. In addition, we applied our method to illustrate the relationship between a dengue fever outbreak and the disease vector.