A method of producing an approximation sequence to a series of sample values, the method comprising the steps of (a) determining a first set having candidate partial sequences as members, each member comprising a plurality of elements; (b) selecting the first n elements of one of the members of the first set as a next output element for said approximation sequence; n a positive integer; (c) forming a second set having descendent candidate partial sequences as members from said first set; (d) applying a fitness filtering process to said second set to rank its members according to fitness for representing at least a corresponding portion of the series of input samples; (e) selecting at least some of the members of the second set to form a third set; and repeating steps (a) to (e) so as to produce said approximation sequence, wherein the third set of step (e) functions as the first set of the subsequent step (a).