xhdr2set

数据挖掘中的关联规则算法

#!/bin/sh
gawk -v app=$3 -v out=$4 '
BEGIN {
  FS = " ";
  if ((app != "") && (out != "")) {
    getline dflt < app;
    while ((getline < app) > 0)
      base[$1] = $2;
  }
  FS = ",";
}
(NR == 1) {
  for (i = 0; ++i <= NF; )
    items[i] = $i;
}
(NR > 1) {
  for (i = k = 0; ++i <= NF; ) {
    if (k++ > 0) printf(" ");
    item = (items[i] "=" $i);
    printf("%s", item);
    if (items[i] in base) apps[item] = base[items[i]];
    else                  apps[item] = dflt;
  }
  printf("\n");
}
END {
  if (out != "") {
    print dflt > out;
    for (t in apps)
      if (apps[t] != dflt)
        printf("%s %s\n", t, apps[t]) > out;
  }
}' $1 > $2