Test of Significance Based on Wavelet Thresholding and Neyman's Truncation

摘要: Abstract Traditional nonparametric tests, such as the Kolmogorov—Smirnov test and Cramer—Von Mises test, are based on empirical distribution functions. Although these tests possess root-n consistency, they effectively use only information contained in low frequencies. This leads to power detecting fine features sharp short aberrants well global high-frequency alternations. The drawback can be repaired via smoothing-based statistics. In this article we propose two kind of statistics wavelet thresholding Neyman truncation. We provide extensive evidence demonstrate that proposed have higher peaks high frequency alternations, while maintaining same capability smooth alternative densities traditional tests. Similar conclusions made for two-sample case, linear rank th...