Statistical inference based on the top scores

In this paper, three different types of data sets, namely, the sequence of top k scores vectors, just one top k scores vector and the sequence of upper k-record values are considered. Frequentist estimation based on three mentioned data sets and also the amount of Fisher information (FI) in them are discussed for a general class of lifetime model, i.e., proportional hazard rate model. It is well-known that this class includes several lifetime distributions such as exponential, Weibull (one parameter), Pareto, Burr type XII and so on. Finally, the two-parameter Weibull distribution is studied in detail. Copyright Sapienza Università di Roma 2012