Nonparametric Estimation on Regression Coefficient and Population Size forIncomplete or Skewed Data

[摘要] Long-tail and skewed data are frequently encountered in botheconomics and health-care fieOds. Since the data on the distribution tailare scarce yet very important, inappropriately handling of such datacan lead to unstable and biased estimation results. We have developeda series of methods to analyze such data. Particularly, we developed amethod to estimate the transformation function and error distributionfunction so to avoid the dLٹcuOt\ of specifying these functions inmodeling. In addition, invoking jointly the smoothing technology,penalty and rank correlation, we have developed a new dimension-freecalculation method to quickly select the important risk factors from alarge number of potential risk factors. Furthermore, we proposed asemi-parametric latent transformation model to combine multipleskewed and long-tail outcomes in a data-driven way. Нe analysis ofreal data showed that our methods are more eٹcLent and robust thanthe existing methods to identify LnfluentLDO risk factors.