統計学輪講（第7回）


日時	2018年5月22日（火） 15時45分～ 16時35分
場所	経済学研究科棟 3階第3教室
講演者	大田浩史（経済学研究科D1）
演題	Semi-parametric estimation of modal regression function and its asymptotic properties
概要	Modal regression estimates the conditional mode of an outcome Y given regressors X. Compared to mean regression, modal regression has particularly useful features when data distribution is highly skewed or has fat tails. In this talk, I will discuss a single index model for modal regression function, and propose a novel estimator of its weighted average derivatives. I construct the estimator based on sample-splitting/cross-fitting techniques. (cf. [1], [2]). In estimating low-dimensional parameters of interest, these approaches is expected to reduce “own-estimating bias ” from estimating highly complex nuisance parameters in full sample. I also derive some asymptotic results of the estimator and valid inference procedures based on sample-splitting/cross-fitting. References [1] V. Chernozhukov, et al. (2017+), “Double/De-Biased Machine Learning for Treatment and Causal Parameters ”, Econometrics Journal. [2] W. Newey and J. Robins (2018), “Cross-Fitting and Fast Remainder Rates for Semiparametric Estimation ”, arXiv:1801.09138.