Revisiting the Evaluation of Robust Regression Techniques for Crop Yield Data Detrending
通过蒙特卡洛实验比较普通最小二乘法和MM估计量在作物产量去趋势中的表现，发现MM估计量在数据受异常值污染时明显优于OLS，建议重新考虑稳健回归方法。
Abstract Using a Monte Carlo experiment, the performance of the ordinary least squares (OLS) and the MM‐estimator, a robust regression technique, is compared in an application of crop yield detrending. Assuming symmetric as well as skewed crop yield distributions, we show that the MM‐estimator performs similarly to OLS for uncontaminated time series of crop yield data, and clearly outperforms OLS for outlier‐contaminated samples. In contrast to earlier studies, our analysis suggests that robust regression techniques, such as the MM‐estimator, should be reconsidered for detrending crop yield data.