Efficient and accurate inference for mixtures of Mallows models with Spearman distance

The Mallows model (MM) occupies a central role in parametric modelling of ranking data to learn preferences population judges. Despite the wide range metrics for rankings that can be considered specification, choice is typically limited Kendall, Cayley or Hamming distances, due closed-form expression related normalizing constant. This work instead focuses on with Spearman distance (MMS). A novel approximation constant introduced allow inference even large number items. allows us develop and implement an efficient accurate EM algorithm estimating finite mixtures MMS aimed at i) enlarging applicability samples drawn from heterogeneous populations, ii) dealing partial affected by diverse forms censoring. These novelties encompass critical inferential steps traditionally use this practice, render comparable (or preferable) MMs other terms computational burden. ability scheme effectiveness are assessed extensive simulation studies. Finally, we show application three real-world datasets endorses our proposals also comparison competing models.