Communication-efficient distributed M-estimation with missing data

Abstract In the big data era, practical applications often encounter incomplete data. Current distributed methods, ignoring missingness, may cause inconsistent estimates. Motivated by that, a distributed algorithm is developed for M-estimation with missing data. The proposed algorithm is communication-efficient, where only gradient information is transferred to the central machine. The parameters of interest and the nuisance parameters are simultaneously updated. Theoretically, it is shown that the proposed algorithm achieves a full sample performance after a moderate number of iterations. The influence of nuisance parameters on distributed M-estimation is also investigated. Simulations via synthetic data illustrate the effectiveness of the algorithm. At last, the algorithm is applied to a real data set.