# gawk profile, created Tue Apr 13 08:47:56 2010

	# BEGIN block(s)

	BEGIN {
     1  	Max = 1
     1  	Seed = 1
     1  	More = 1.02
     1  	Lives = 5
     1  	Dull = 0.1
     1  	Beam = 10
     1  	Samples = 20
     1  	Pinch = 1 / 1000
     1  	OverFitted = 3
     1  	CONVFMT = "%.8g"
     1  	IGNORECASE = 1
     1  	SUBSEP = "="
     1  	_ = SUBSEP
     1  	OFS = ","
     1  	C = ","
     1  	Verbose = 1
	}

	# Rule(s)

   243  /@attribute/	{ # 13
     5  	Name[++Name[0]] = $2
     5  	Name[$2] = Name[0]
	}

   243  {
   243  	gsub(/[ \\t]*/, "", $0)
	}

   243  {
   243  	gsub(/%.*/, "", $0)
	}

   243  /^$/	{ # 141
    69  	next
	}

   174  /@data/	{ # 5
     1  	In = 1
     1  	FS = ","
     1  	srand(Seed)
	}

   174  /@/	{ # 17
     7  	next
	}

   167  In	{ # 167
   167  	Rows++
   167  	train(Rows, Data, Counts)
	}

	# END block(s)

	END {
     1  	learn(Rows, Data, Counts)
	}

	# Functions, listed alphabetically

    10  function chop(score0, memo, out, score, n, i)
	{
    10  	n = asort(score0, score)
   110  	for (i = n; i >= 1; i--) {
   110  		if (score[i] <= score[n] * Dull) {
				break
			}
   110  		if (i <= n - Beam) { # 10
    10  			break
			}
   100  		out[memo[score[i]]] = score[i]
		}
	}

   180  function combine(class, this, that, n, i, used, tmp, out)
	{
   180  	sub(/^[^,]*,/, "", this)
   180  	sub(/^[^,]*,/, "", that)
   180  	split(((this) "," (that)), tmp, ",")
   180  	n = asort(tmp)
   180  	out = tmp[1]
   180  	used[tmp[1]] = 1
   694  	for (i = 1; i <= n; i++) {
   694  		if (! used[tmp[i]]) { # 414
   414  			out = out "," tmp[i]
   414  			used[tmp[i]] = 1
			}
		}
   180  	return ((class) "," (out))
	}

     9  function explode(a, out, i, j)
	{
    90  	for (i in a) {
   866  		for (j = 1; j <= a[i]; j++) {
   866  			out[++out[0]] = i
			}
		}
	}

	function fv(f, v)
	{
		return ((f) _ (v))
	}

     1  function learn(rows, data, counts, which0, which)
	{
     1  	round0(counts, which0)
     1  	rounds(1, 0, Lives, which0, rows, data, which)
	}

  8851  function matched(row, data, this, out, col, n, goals, pair, f0, f, status)
	{
  8851  	n = split(this, goals, ",")
 19536  	for (col = 2; col <= n; col++) {
 19536  		split(goals[col], pair, _)
 19536  		f = pair[1]
 19536  		status[f] += data[row, f] == pair[2]
 19536  		if (f0 && f != f0 && ! status[f0]) { # 4982
  4982  			return 0
			}
 14554  		f0 = f
		}
  3869  	return status[f]
	}

     9  function most(a, i, max)
	{
     9  	max = -1e+09
    90  	for (i in a) {
    90  		if (a[i] > max) { # 19
    19  			max = a[i]
			}
		}
     9  	return max
	}

     9  function normalize(a, i, sum)
	{
    90  	for (i in a) {
    90  		sum += a[i]
		}
    90  	for (i in a) {
    90  		a[i] = int(100 * a[i] / sum)
		}
	}

   416  function one(sample, any)
	{
   416  	any = int(rand() * sample[0]) + 1
   416  	return sample[any]
	}

     9  function o(a, s, what, i, com)
	{
     9  	print ""
     9  	com = what ? ("sort " (what)) : "sort  "
    90  	for (i in a) {
    90  		print((a[i] "," (i))) | (com)
		}
     9  	close(com)
	}

	function oo(a, s, what, i, com)
	{
		print ""
		com = what ? ("sort  " (what)) : "sort   "
		for (i in a) {
			print(((s) "[ " (i) " ] = " a[i])) | (com)
		}
		close(com)
	}

     1  function round0(counts, which, j, i, s, memo, score)
	{
    70  	for (i in counts) {
    70  		s = Max ? counts[i] : -1 * counts[i]
    70  		j = (Max "," (i))
    70  		memo[s] = j
    70  		score[j] = s
		}
     1  	chop(score, memo, which)
	}

    10  function rounds(round, max0, lives, which0, rows, data, out, score, max, i, sample, which1, s, memo, which2)
	{
    10  	if (round == 1) { # 1
     1  		max = 0
     9  	} else {
     9  		max = most(which0)
     9  		lives = max > max0 * More ? Lives : lives - 1
     9  		if (lives < 0) { # 1
    10  			for (i in which0) {
    10  				out[i] = which0[i]
				}
     1  			return max
			}
		}
     9  	print "\n----------------------------------------------------"
     9  	print ("% max: " (max) " seed: " Seed " round: " (round) " max: " (max) " lives: " (lives))
     9  	normalize(which0)
     9  	explode(which0, sample)
     9  	twos(Max, sample, Samples, which1)
    90  	for (i in which0) {
    90  		which1[i] = i
		}
     9  	if (Verbose) { # 9
     9  		values(which1, "candidate")
		}
   166  	for (i in which1) {
   166  		s = i in score ? score[i] : score(i, rows, data) + rand() * Pinch
   166  		memo[s] = i
   166  		score[i] = s
		}
     9  	chop(score, memo, which2)
     9  	if (Verbose) { # 9
     9  		o(which2, "score", "-t, -n -k 1")
		}
     9  	return rounds(round + 1, max, lives, which2, rows, data, out, score)
	}

    53  function score1(fits, support, sum)
	{
    53  	if (fits <= OverFitted) {
			return 0
		}
    53  	if (Eval == 1) {
			return support * sum
		}
    53  	return sum
	}

    53  function score(rule, rows, data, cols, row, s, triggered, fits, sum, support)
	{
    53  	fits = sum = Pinch
    53  	cols = Name[Name[0]]
  8851  	for (row = 1; row <= rows; row++) {
  8851  		if (matched(row, data, rule)) { # 1333
  1333  			fits++
  1333  			sum += data[row, cols]
			}
		}
    53  	support = fits / rows
    53  	s = score1(fits, support, sum)
    53  	return Max ? s : -1 * s
	}

   167  function train(rows, d, c, what, i)
	{
   835  	for (i = 1; i <= NF; i++) {
   835  		if ($(i) == "?") {
				continue
			}
   835  		what = Name[i]
   835  		d[rows, what] = $(i)
   835  		c[what, $(i)] += $NF
		}
	}

     9  function twos(class, sample, n, sampled, pair)
	{
   180  	while (n--) {
   180  		pair = two(class, sample)
   180  		sampled[pair] = pair
		}
	}

   208  function two(class, sample, tries, this, that)
	{
   208  	this = one(sample)
   208  	if (tries == 9) {
			return this
		}
   208  	that = one(sample)
   208  	if (this == that) { # 28
    28  		return two(class, sample, tries + 1)
   180  	} else {
   180  		return combine(class, this, that)
		}
	}

     9  function values(a, s, what, i, com)
	{
     9  	print ""
     9  	com = what ? ("sort " (what)) : "sort"
   166  	for (i in a) {
   166  		print(("% " (s) ": " a[i])) | (com)
		}
     9  	close(com)
	}