Discussion of “ Location-Scale Depth ”

@inproceedings{Mller2004DiscussionO,
  title={Discussion of “ Location-Scale Depth ”},
  author={Ch. H. M{\"u}ller},
  year={2004}
}
Mizera’s previous work [13] introduced tangent depth, a powerful method for defining robust statistics. In this framework, one measures the quality of a putative fit θ to a sample point zi by a criterial function Fi(θ). We say that θ is a nonfit if there is another fit φ with Fi(φ) < Fi(θ) for all i. The depth of θ is then defined as the smallest value d such that d of the sample points can be removed causing θ to be a nonfit for the remaining data. This method actually leads to two notions of… CONTINUE READING