A FORTRAN IV program for comparing ranking algorithms in quantitative biostratigraphy

摘要: Abstract A program for evaluating the performance of competing ranking algorithms in stratigraphic paleontology is presented. The (1) generates a hypothetical, and thus known, succession taxa time (2) simulates their strata at several local sample sites. If desired, may be repeated (=50 or 100 example) iterations site data each sent to two user routines inferred rankings (inferred events time). First first last occurrences (fads lads) taken together, then lads-only, fads-only sent. For submission routine, Kendall rank correlation coefficients Spearman are computed comparing generated by routine with known time. compared obtaining submitted dataset differences between corresponding (and/or Spearman) algorithms, testing observed statistical significance. simple two-sided t -test used test whether mean difference differs significantly from zero; if c -tests performed, level significance should set alpha/ obtain maximum experimentwise error rate less than alpha. compare three provided Agterberg Nel (1982a, b) as well determine work datasets combining lads fads vs lads-only fads-only. Nel's Presorting algorithm performed better Ranking Scaling algorithm. All slightly but on opposed combined data.