djc_prep

来自「数据挖掘中的一算法 ines算法 c下实现的。适合初学习数据挖掘者借鉴」· 代码 · 共 56 行

#!/bin/sh
#-----------------------------------------------------------------------
# File    : djc_prep
# Contents: preparation of Danish Jersey Cattle data
# Author  : Christian Borgelt
# History : ??.??.1999 file created
#-----------------------------------------------------------------------

gawk '
BEGIN {
  # printf "parental_error "
  printf "dam_correct sire_correct "
  printf "true_dam_pg1 true_dam_pg2 true_sire_pg1 true_sire_pg2 "
  printf "stated_dam_pg1 stated_dam_pg2 stated_sire_pg1 stated_sire_pg2 "
  printf "offspring_pg1 offspring_pg2 "
  printf "offspring_genotype "
  printf "factor_40 factor_41 factor_42 factor_43 "
  printf "lysis_40 lysis_41 lysis_42 lysis_43\n"
}
{
  # --- parental error, dam error, sire error
  # printf "%s ", $16
  if   ($16 == "n") printf "y y "
  else              printf "* * "

  # --- phenogroups of true dam
  printf "%s %s ", $12, $13

  # --- phenogroups of true sire
  printf "%s %s ", $14, $15

  # --- phenogroups of stated dam
  printf "%s %s ", $5, $6

  # --- phenogroups of stated sire
  printf "%s %s ", $7, $8

  # --- phenogroups of offspring
  printf "%s %s ", $10, $11

  # --- genotype of offspring
  printf "%s ", $9

  # --- factors 40,...,43
  if      ($9 == "f1f1") printf "y y n n "
  else if ($9 == "f1v1") printf "y y y y "
  else if ($9 == "f1v2") printf "y y n y "
  else if ($9 == "v1v1") printf "n n y y "
  else if ($9 == "v1v2") printf "n y y y "
  else if ($9 == "v2v2") printf "n y n y "
  else                   printf "* * * * "

  # --- lysis 40,...,43
  printf "%s %s %s %s\n", $1, $2, $3, $4
}' djc.raw > djc.tab