A0693
Title: Testing heteroskedasticity in high-dimensional linear regression
Authors: Akira Shinkyu - Shiga University (Japan) [presenting]
Abstract: A new testing procedure of heteroskedasticity is proposed in high-dimensional linear regression, where the number of covariates can be larger than the sample size. The testing procedure is based on residuals of the Lasso. The test statistic is demonstrated to have asymptotic normality under the null hypothesis of homoscedasticity. Simulation results show that the proposed testing procedure obtains accurate empirical sizes and powers.
