--- /dev/null
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require "rubygems"
+require "mcmd"
+
+require "lcm"
+require "zdd"
+require "traDB.rb"
+
+module TAKE
+
+
+#========================================================================
+# 列挙関数:lcm 利用DB:TraDB
+#========================================================================
+class LcmIs
+ attr_reader :size # 列挙されたパターン数
+
+ def reduceTaxo(pat,items)
+ tf=MCMD::Mtemp.new
+
+ if items.taxonomy==nil then
+ return pat
+ end
+
+ xxrt = tf.file
+ taxo=items.taxonomy
+ f=""
+ f << "mtrafld f=#{taxo.itemFN},#{taxo.taxoFN} -valOnly a=__fld i=#{taxo.file} o=#{xxrt}"
+ system(f)
+
+ # xxrtの内容：oyakoに親子関係にあるアイテム集合のリストが格納される
+ # __fld
+ # A X
+ # B X
+ # C Y
+ # D Z
+ # E Z
+ # F Z
+ oyako=ZDD.constant(0)
+ MCMD::Mcsvin.new("i=#{xxrt}"){|csv|
+ csv.each{|fldVal|
+ items=fldVal["__fld"]
+ oyako=oyako+ZDD.itemset(items)
+ }
+ }
+
+ # 親子リストにあるアイテム集合を含むパターンを削除する
+ pat=pat.restrict(oyako).iif(0,pat)
+
+ return pat
+ end
+
+ def initialize(db)
+ @temp=MCMD::Mtemp.new
+
+ @db = db # 入力データベース
+
+ @file=@temp.file
+
+ items=@db.items
+
+ # アイテムをシンボルから番号に変換する。
+ f=""
+ f << "msortf f=#{@db.itemFN} i=#{@db.file} |"
+ f << "mjoin k=#{@db.itemFN} K=#{items.itemFN} m=#{items.file} f=#{items.idFN} |"
+ f << "mcut f=#{@db.idFN},#{items.idFN} |"
+ f << "msortf f=#{@db.idFN} |"
+ f << "mtra k=#{@db.idFN} f=#{items.idFN} |"
+ f << "mcut f=#{items.idFN} -nfno o=#{@file}"
+ system(f)
+ end
+
+ def enumerate(type, minSup, lenLB=1, lenUB=4, top=10000, minSupCnt=0)
+
+ tf=MCMD::Mtemp.new
+
+ @type = type
+ if minSupCnt!=0 then
+ @minSupCnt = minSupCnt
+ @minSup = minSupCnt.to_f / @db.size.to_f
+ else
+ @minSup = minSup.to_f
+ @minSupCnt = (@minSup * @db.size.to_f + 0.99).to_i
+ end
+ @lenLB = lenLB
+ @lenUB = lenUB
+ @top = top
+
+ xxp = tf.file #MCMD::Mtemp.new
+ xxt = tf.file #MCMD::Mtemp.new
+
+ if(top==0) then
+ MCMD::lcm("type=#{@type} i=#{@file} s=#{@minSupCnt} l=#{@lenLB} u=#{@lenUB} o=#{xxp} t=#{xxt}")
+ else
+ MCMD::lcm("type=#{@type} i=#{@file} s=#{@minSupCnt} l=#{@lenLB} u=#{@lenUB} o=#{xxp} t=#{xxt} K=#{@top}")
+ end
+
+ # パターンのサポートを計算しCSV出力する
+ MCMD::msgLog("output patterns to CSV file ...")
+ xxp0=tf.file
+ @pFile = @temp.file
+ items=@db.items
+ f=""
+ f << "mcut -nfni f=0:pid,1:pattern,2:count i=#{xxp} |"
+ f << "mdelnull f=pattern |"
+ f << "mvreplace vf=pattern m=#{items.file} K=#{items.idFN} f=#{items.itemFN} |"
+ f << "msetstr v=#{@db.size} a=total |" # トータル件数
+ f << "mcal c='${count}/${total}' a=support |" # サポートの計算
+ f << "mcut f=pid,pattern,count,total,support |"
+ f << "mtra -r f=pattern |"
+ f << "msortf f=pid,pattern |"
+ f << "mtra k=pid f=pattern |"
+ f << "mvsort vf=pattern |"
+ f << "msortf f=pattern o=#{xxp0}"
+ system(f)
+
+ # taxonomy指定がない場合(2010/11/20追加)
+ if items.taxonomy==nil then
+ FileUtils.cp(xxp0, @pFile)
+
+ # taxonomy指定がある場合
+ else
+ MCMD::msgLog("reducing redundant rules in terms of taxonomy ...")
+ zdd=ZDD.constant(0)
+ MCMD::Mcsvin.new("i=#{xxp0}"){|csv|
+ csv.each{|fldVal|
+ items=fldVal['pattern']
+ zdd=zdd+ZDD.itemset(items)
+ }
+ }
+
+ zdd=reduceTaxo(zdd,@db.items)
+ xxp1=tf.file
+ xxp2=tf.file
+ zdd.csvout(xxp1)
+ f=""
+ f << "mcut -nfni f=1:pattern i=#{xxp1} |"
+ f << "mvsort vf=pattern |"
+ f << "msortf f=pattern o=#{xxp2}"
+ system(f)
+
+ f=""
+ f << "msortf f=pattern i=#{xxp0} |"
+ f << "mcommon k=pattern m=#{xxp2} |"
+ f << "msortf f=support%nr o=#{@pFile}"
+ system(f)
+
+ end
+
+ @size = MCMD::mrecount("i=#{@pFile}") # 列挙されたパターンの数
+ MCMD::msgLog("the number of patterns enumerated is #{@size}")
+
+ # トランザクション毎に出現するシーケンスを書き出す
+ MCMD::msgLog("output tid-patterns ...")
+ @tFile = @temp.file
+
+ xxp3=tf.file
+ f=""
+ f << "mcut f=#{@db.idFN} i=#{@db.file} |"
+ f << "muniq k=#{@db.idFN} |"
+ f << "mnumber S=0 a=__tno |"
+ f << "msortf f=__tno o=#{xxp3}"
+ system(f)
+
+ xxp4=tf.file
+ f=""
+ f << "mcut f=pid i=#{@pFile} |"
+ f << "msortf f=pid o=#{xxp4}"
+ system(f)
+
+ f=""
+ f << "mcut -nfni f=0:__tno,1:pid i=#{xxt} |"
+ f << "msortf f=pid |"
+ f << "mcommon k=pid m=#{xxp4} |"
+ f << "msortf f=__tno |"
+ f << "mjoin k=__tno m=#{xxp3} f=#{@db.idFN} |"
+ f << "mcut f=#{@db.idFN},pid o=#{@tFile}"
+ system(f)
+ end
+
+ def output(outpath)
+ system "mv #{@pFile} #{outpath}/patterns.csv"
+ system "mv #{@tFile} #{outpath}/tid_pats.csv"
+ end
+end
+
+end #module
+
OSDN Git Service
