محمود افشاری

The records in a series of observations are those values which are more or less than their previous values. The minimization of the number of the records in an arbitrary sample as well as the infinity of the expected average time for occurrence of a new record is two main characteristics of records. These two foresaid specifications lead to time taking simulation process of the records from one hand and from the other hand the size of the record samples in statistical inference based on record values to be minimized. In this paper, proposing record extraction methods, an algorithm will be presented which produce the upper (lower) records from a continuous distribution in a short time.