def g2pair(self,ni,nf,ei,ef1,ef2,ew,numFile,mapFile,weightFile):
inobj = []
inobj.append(nm.mcut(f="%s:node"%(ef1),i=ei ).msetstr(a="flag",v=0))
inobj.append(nm.mcut(f="%s:node"%(ef2),i=ei ).msetstr(a="flag",v=0))
if nf :
inobj.append(nm.mcut(f="%s:node"%(nf),i=ni ).msetstr(a="flag",v=1))
f = nm.mbest(i=inobj,k="node",s="flag",fr=0,size=1 )
# isolated nodes are set to the end of position in mapping file.
# S= must start from 0 (but inside R vertex number will be added one)
f <<= nm.mnumber(s="flag,node",a="num",S=0,o=mapFile)
f.run()
f = None
f <<= nm.mcut(f=[ef1,ef2] , i=ei)
f <<= nm.mjoin( k=ef1 , K="node" , m=mapFile ,f="num:num1")
f <<= nm.mjoin( k=ef2 , K="node" , m=mapFile ,f="num:num2")
f <<= nm.mcut(f="num1,num2")
f <<= nm.mfsort(f="num1,num2")
f <<= nm.msortf(f="num1%n,num2%n",nfno=True)
f <<= nm.cmd("tr ',' ' ' " )
f <<= nm.mwrite(o=numFile)
f.run()
nodeSize=mrecount(i=mapFile)
if ew:
nm.mcut(f=ew,i=ei,o=weightFile).run()
else:
ew="weight"
nm.msetstr(v=1,a=ew,i=ei).mcut(f=ew,o=weightFile).run()
return nodeSize
def runmain(self, edgeFile):
import re
baseName = re.sub('\.edge$', "", edgeFile)
name = re.sub('^.*\/', "", baseName)
nodeFile = re.sub('\.edge$', ".node", edgeFile)
# convert the original graph to one igraph can handle
temp = Mtemp()
xxnum = temp.file()
xxmap = temp.file()
xxout = temp.file()
xxscp = temp.file()
nodeSize = self.g2pair(nodeFile, self.nf, edgeFile, self.ef1, self.ef2,
xxnum, xxmap)
# generate R script, and run
self.genRscript(self.directed, self.pars, xxnum, nodeSize, xxout,
xxscp)
if self.verbose:
os.system("R --vanilla -q < %s" % (xxscp))
else:
os.system("R --vanilla -q --slave < %s 2>/dev/null " % (xxscp))
nm.msetstr(v=name, a="id",
i=xxout).mcut(x=True,
f="0L,0-1L",
o=self.oPath + "/" + name + ".csv").run()
def __mkEdge(key, ef1, ef2, el, ec, ed, ev, ei, norm, mapFile, oFile):
# mcal cat用のlabel項目の作成
label = []
if el:
for nml in el:
label.append("$s{" + nml + "}")
evcdStr = []
#エッジの拡大率は固定
er = 10
if ev:
evcdStr.append(ev + ":ev")
if ec:
evcdStr.append(ec + ":ec")
if ed:
evcdStr.append(ed + ":ed")
f = None
if el:
f <<= nm.mcal(c='cat(\"_\",%s)' % (','.join(label)), a="##label", i=ei)
else:
f <<= nm.msetstr(v="", a="##label", i=ei)
if len(evcdStr) == 0:
f <<= nm.mcut(f="%s:key,%s:nam1,%s:nam2,##label:el" % (key, ef1, ef2))
else:
f <<= nm.mcut(f="%s:key,%s:nam1,%s:nam2,##label:el,%s" %
(key, ef1, ef2, ','.join(evcdStr)))
if not ev:
f <<= nm.msetstr(v="", a="ev")
if not ed:
f <<= nm.msetstr(v="", a="ed")
if not ec:
f <<= nm.msetstr(v="", a="ec")
f <<= nm.mnullto(f="key", v="##NULL##")
f <<= nm.mjoin(k="key", K="nam", m=mapFile, f="num:keyNum")
f <<= nm.mjoin(k="nam1", K="nam", m=mapFile, f="num:num1,leaf:leaf1")
f <<= nm.mjoin(k="nam2", K="nam", m=mapFile, f="num:num2,leaf:leaf2")
if norm:
f <<= nm.mnormalize(f="ev:ev2", c="range")
f <<= nm.mcal(c='${ev2}*(%s-1)+1' % (er), a="evv")
f <<= nm.mcut(
f="key,nam1,nam2,keyNum,num1,num2,el,evv:ev,ed,ec,leaf1,leaf2",
o=oFile)
else:
f <<= nm.mcut(
f="key,nam1,nam2,keyNum,num1,num2,el,ev,ed,ec,leaf1,leaf2",
o=oFile)
f.run()
def __mkMap(key, nf, ni, ef1, ef2, ei, oFile):
# leaf nodeの構築
infL = [
nm.mcommon(k=ef1, K=key, m=ei, r=True, i=ei).mcut(f="%s:nam" % (ef1)),
nm.mcommon(k=ef2, K=key, m=ei, r=True, i=ei).mcut(f="%s:nam" % (ef2))
]
if ni:
infL.append(
nm.mcommon(k=nf, K=key, m=ei, r=True,
i=ni).mcut(f="%s:nam" % (nf)))
xleaf = nm.muniq(i=infL, k="nam")
xleaf <<= nm.msetstr(v=1, a="leaf")
if ni:
inp = [
nm.mcut(f="%s:nam" % (nf), i=ni),
nm.mcut(f="%s:nam" % (key), i=ni)
]
else:
inp = [
nm.mcut(f="%s:nam" % (ef1), i=ei),
nm.mcut(f="%s:nam" % (ef2), i=ei),
nm.mcut(f="%s:nam" % (key), i=ei)
]
f = None
f <<= nm.muniq(k="nam", i=inp)
f <<= nm.mjoin(k="nam", m=xleaf, f="leaf", n=True)
# nullは最初に来るはずなので、mcalでなくmnumberでもnullを0に採番できるはずだが念のために
f <<= nm.mcal(c='if(isnull($s{nam}),0,line()+1)', a="num")
f <<= nm.mnullto(f="nam", v="##NULL##", o=oFile)
f.run()
def g2pair(self,ni,nf,ei,ef1,ef2,ew,numFile,mapFile,weightFile):
#MCMD::msgLog("converting graph files into a pair of numbered nodes ...")
#wf=MCMD::Mtemp.new
#wf1=wf.file
#wf2=wf.file
#wf3=wf.file
allinObj =[]
wf1 = nm.mcut(f="%s:node"%(ef1),i=ei ).msetstr(v=0,a="flag")
wf2 = nm.mcut(f="%s:node"%(ef2),i=ei ).msetstr(v=0,a="flag")
f = None
if nf:
f <<= nm.mcut(i=[wf1,wf2,nm.mcut(f=nf+":node",i=ni).msetstr(v=1,a="flag")],f="node,flag")
f <<= nm.mbest(k="node" , s="flag" , fr=0 ,size=1)
else:
f <<= nm.mcut(i=[wf1,wf2],f="node,flag")
f <<= nm.muniq(k="node")
f <<= nm.mnumber(s="flag,node",a="num",S=0,o=mapFile)
f.run()
f = None
f <<= nm.mcut(f=[ef1,ef2],i=ei)
f <<= nm.mjoin(k=ef1 , K="node",m=mapFile , f="num:num1")
f <<= nm.mjoin(k=ef2 , K="node",m=mapFile , f="num:num2")
f <<= nm.mcut(f="num1,num2")
f <<= nm.mfsort(f="num1,num2")
f <<= nm.msortf(f="num1%n,num2%n",nfno=True)
f <<= nm.cmd("tr ',' ' ' ")
f <<= nm.mwrite(o=numFile)
f.run()
if ew :
nm.mcut(f=ew,i=ei,o=weightFile).run()
else:
ew="weight"
nm.msetstr(v=1 , a=ew ,i=ei).mcut(f=ew,o=weightFile).run()
nodeSize=mrecount(i=mapFile)
return nodeSize
def mnest2tree(ei, ef, k, ni=None, nf=None, ev=None, no=None, eo=None):
# paracheck追加
efs = ef.split(",")
ef1 = efs[0]
ef2 = efs[1]
f = nm.mcut(f="%s:#orgKey,%s:#orgEf1,%s:#orgEf2" % (k, ef1, ef2), i=ei)
temp = mtemp.Mtemp()
of = temp.file()
with _nu.mcsvout(o=of, f="#orgKey,#orgEf1,#orgEf2,#ef1,#ef2") as oCSV:
for flds in f:
orgKey = flds[0]
orgEf1 = flds[1]
orgEf2 = flds[2]
oCSV.write([orgKey, orgEf1, orgEf2, orgKey, orgEf1])
oCSV.write([orgKey, orgEf1, orgEf2, orgKey, orgEf2])
f = None
f <<= nm.mjoin(k="#orgKey,#orgEf1,#orgEf2", K=[k, ef1, ef2], m=ei,
i=of) # 全項目join
if ev:
f <<= nm.mavg(k="#ef1,#ef2", f=ev)
else:
f <<= nm.muniq(k="#ef1,#ef2")
f <<= nm.mcut(r=True, f="#orgKey,#orgEf1,#orgEf2")
f <<= nm.mfldname(f="#ef1:%s,#ef2:%s" % (ef1, ef2), o=eo)
f.run()
if ni:
head = nu.mheader(i=ni)
fldnames = [s for s in head if s != nf]
commas = ',' * (len(fldnames) - 1)
f0 = None
f0 <<= nm.mcut(f="%s:%s" % (ef1, nf), i=eo)
f0 <<= nm.muniq(k=nf)
f0 <<= nm.mcommon(k=nf, m=ni, r=True)
f0 <<= nm.msetstr(v=commas, a=fldnames)
f = nm.mcut(f=k, r=True, i=[ni, f0])
f <<= nm.msetstr(v="", a=k, o=no)
f.run()
def tra2tbl(iFile,
oFile,
tidFld,
itemFld,
null=0,
dummy=True,
aggFld=None,
aggStat=None,
klassFld=None):
klass = None
if klassFld is not None and klassFld != "":
klass <<= nm.mcut(f=tidFld + "," + klassFld, i=iFile)
klass <<= nm.muniq(k=tidFld)
f = None
# aggFldが指定されていればセル項目を集計
if aggFld is not None and aggFld != "":
f <<= nm.mcut(f=tidFld + "," + itemFld + "," + aggFld, i=iFile)
f <<= nm.mstats(k=tidFld, f="%s:_cell" % (aggFld), c=aggStat)
# aggFldが指定されていなければカウント
else:
f <<= nm.mcut(f=tidFld + "," + itemFld, i=iFile)
# アイテムが出現したかどうか
if dummy:
f <<= nm.muniq(k=tidFld + "," + itemFld)
f <<= nm.msetstr(v=1, a="_cell")
# アイテムが何件出現したかどうか
else:
f <<= nm.mcount(k=tidFld + "," + itemFld, a="_cell")
# 横展開
f <<= nm.m2cross(k=tidFld, s=itemFld, f="_cell")
# クラス項目が指定されていれば結合する
if klassFld is not None and klassFld != "":
f <<= nm.mjoin(k=tidFld, m=klass, f=klassFld)
# null値を一斉に置換する
f <<= nm.mnullto(f="*", v=null, o=oFile)
f.run(msg="on")
def run(self):
temp = mtemp.Mtemp()
### mtra2gc
xxsimgN = temp.file()
xxsimgE = temp.file()
xxsimgE0 = temp.file()
param = {}
param["i"] = self.iFile
if self.idFN:
param["tid"] = self.idFN
if self.itemFN:
param["item"] = self.itemFN
if self.sp1:
param["s"] = self.sp1
if self.sp2:
param["S"] = self.sp2
#####################
# 異なる向きのconfidenceを列挙するためにsim=C th=0として双方向列挙しておく
# 出力データは倍になるが、mfriendsで-directedとすることで元が取れている
param["sim"] = "C"
param["th"] = "0"
param["node_support"] = True
if self.numtp:
param["num"] = True
param["no"] = xxsimgN
param["eo"] = xxsimgE0
nt.mtra2gc(**param).run()
f = nm.readcsv(xxsimgE0)
for i in range(self.filterSize):
f <<= nm.mselnum(f=self.filter[i],
c="[%s,%s]" % (self.lb[i], self.ub[i]))
f <<= nm.writecsv(xxsimgE)
f.run()
### mfrirends
xxfriends = temp.file()
xxfriendE = temp.file()
xxw = temp.file()
xxf = temp.file()
xxff = temp.file()
xxor = temp.file()
if not os.path.isdir(xxfriends):
os.makedirs(xxfriends)
col = [["FF000080", "FF888880"], ["0000FF80", "8888FF80"],
["00FF0080", "88FF8880"]]
for i in range(len(self.sim)):
paramf = {}
paramf["ei"] = xxsimgE
paramf["ni"] = xxsimgN
paramf["ef"] = "node1,node2"
paramf["nf"] = "node"
paramf["eo"] = xxfriendE
paramf["no"] = xxfriends + "/n_" + str(i)
paramf["sim"] = self.sim[i]
paramf["dir"] = self.dir[i]
paramf["rank"] = self.rank[i]
paramf["directed"] = True
nt.mfriends(**paramf).run()
frec2 = nm.mfsort(f="node1,node2", i=xxfriendE)
frec2 <<= nm.msummary(k="node1,node2",
f=self.sim[i],
c="count,mean")
frec2 <<= nm.mselstr(f="count", v=2)
# node1%0,node2%1,fld,count,mean
# a,b,support,2,0.1818181818
# a,d,support,2,0.1818181818
f = nm.mjoin(k="node1,node2",
K="node1,node2",
m=frec2,
f="mean:s1",
n=True,
i=xxfriendE)
f <<= nm.mjoin(k="node2,node1",
K="node1,node2",
m=frec2,
f="mean:s2",
n=True)
# 1) xxrecs2でsimをjoinできない(s1,s2共にnull)ということは、それは片方向枝なので"F"をつける
# 2) 双方向枝a->b,b->aのうちa->bのみ(s1がnullでない)に"W"の印をつける。
# 3) それ以外の枝は"D"として削除
f <<= nm.mcal(
c='if(isnull($s{s1}),if(isnull($s{s2}),\"F\",\"D\"),\"W\")',
a="dir")
f <<= nm.mselstr(f="dir", v="D", r=True)
f <<= nm.mcal(c='if($s{dir}==\"W\",$s{s1},$s{%s})' % (self.sim[i]),
a="sim")
f <<= nm.mchgstr(f="dir:color",
c='W:%s,F:%s' % (col[i][0], col[i][1]),
A=True)
f <<= nm.msetstr(v=[self.sim[i], str(i)], a="simType,simPriority")
f <<= nm.mcut(f="simType,simPriority,node1,node2,sim,dir,color",
o=xxfriends + "/e_" + str(i))
f.run()
# node1%1,node2%0,simType,sim,dir,color
# b,a,jaccard,0.3333333333,F,8888FF
# j,c,jaccard,0.3333333333,F,8888FF
# b,d,jaccard,0.3333333333,F,8888FF
# a,e,jaccard,0.5,W,0000FF
# d,e,jaccard,0.5,W,0000FF
# rule fileの出力
if self.orFile:
mmm = nm.mcat(i=xxfriends + "/e_*").muniq(k="node1,node2")
nm.mcommon(k="node1,node2", i=xxsimgE, m=mmm, o=self.orFile).run()
# マルチ枝の単一化(W優先,パラメータ位置優先)
if self.prune:
"""
# 双方向と片方向に分割
nm.mcat(i=xxfriends+"/e_*").mselstr(f="dir",v="W",o=xxw,u=xxf).run()
# 片方向のみの枝を選択
f = nm.mcommon(k="node1,node2",K="node1,node2",r=True,m=xxw,i=xxf)
f <<= nm.mcommon(k="node1,node2",K="node2,node1",r=True,m=xxw,o=xxff)
f.run()
f = nm.mcat(i=xxw+","+xxff).mbest(k="node1,node2",s="dir%r,simPriority%n",o=self.oeFile).run()
"""
#これだめ
fo = nm.mcat(i=xxfriends + "/e_*").mselstr(f="dir", v="W")
fu = fo.direction("u") # これは再考
fu <<= nm.mcommon(k="node1,node2", K="node1,node2", r=True, m=fo)
fu <<= nm.mcommon(k="node1,node2", K="node2,node1", r=True, m=fo)
#f = nm.m2cat()
f = nm.mbest(i=[fo, fu],
k="node1,node2",
s="dir%r,simPriority%n",
o=self.oeFile)
f.run()
else:
nm.mcat(i=xxfriends + "/e_*", o=self.oeFile).run()
nm.mcat(i=xxfriends + "/n_0", o=self.onFile).run()
def enumerate(self, eArgs):
tf = mtemp.Mtemp()
# 最小サポートと最小サポート件数
if "minCnt" in eArgs and eArgs["minCnt"] != None:
self.minCnt = int(eArgs["minCnt"])
self.minSup = float(self.minCnt) / float(self.db.size)
else:
self.minSup = float(eArgs["minSup"])
self.minCnt = int(self.minSup * float(self.db.size) + 0.99)
# 最大サポートと最大サポート件数
self.maxCnt = None
if ("maxCnt" in eArgs
and eArgs["maxCnt"] != None) or ("maxSup" in eArgs
and eArgs["maxSup"] != None):
if "maxCnt" in eArgs and eArgs["maxCnt"] != None:
self.maxCnt = int(eArgs["maxCnt"])
self.maxSup = float(self.maxCnt) / float(self.db.size)
else:
self.maxSup = float(eArgs["maxSup"])
self.maxCnt = int(self.maxSup * float(self.db.size) + 0.99)
# 列挙パターン数上限が指定されれば、一度lcmを実行して最小サポートを得る
if "top" in eArgs and eArgs["top"] != None:
self.top = eArgs["top"]
# 列挙パターン数上限が指定されれば、一度lcmを実行して最小サポートを得る
if self.top and self.top > 0:
xxtop = tf.file()
extTake.lcmseq(type="Cf",
K=str(self.top),
i=self.file,
sup="1",
so=xxtop)
with open(xxtop, "r") as rfile:
self.minCnt = int(rfile.read().strip())
# lcm_seq出力ファイル
lcmout = tf.file()
# 頻出パターンがなかった場合、lcm出力ファイルが生成されないので
# そのときのために空ファイルを生成しておいく。
with open(lcmout, "w") as efile:
pass
# lcm_seqのパラメータ設定と実行
params = {}
if self.msgoff:
params["type"] = "CIf_"
else:
params["type"] = "CIf"
if self.maxCnt:
params["U"] = str(self.maxCnt)
if "minLen" in eArgs:
params["l"] = str(eArgs["minLen"])
if 'maxLen' in eArgs:
params["u"] = str(eArgs["maxLen"])
if 'gap' in eArgs:
params["g"] = str(eArgs["gap"])
if 'win' in eArgs:
params["G"] = str(eArgs["win"])
params["i"] = self.file
params["sup"] = str(self.minCnt)
params["o"] = lcmout
# lcm_seq実行
#MCMD::msgLog("#{run}")
if 'padding' in eArgs and eArgs[
"padding"]: # padding指定時は、0アイテムを出力しないlcm_seqを実行
extTake.lcmseq_zero(params)
else:
extTake.lcmseq(params)
# パターンのサポートを計算しCSV出力する
self.pFile = self.temp.file()
items = self.db.items
transl = self.temp.file()
extTake.lcmtrans(lcmout, "p", transl)
f = nm.mdelnull(f="pattern", i=transl)
f <<= nm.mvreplace(vf="pattern",
m=items.file,
K=items.idFN,
f=items.itemFN)
f <<= nm.msetstr(v=self.db.size, a="total")
f <<= nm.mcal(c='${count}/${total}', a="support") # サポートの計算
f <<= nm.mcut(f="pid,pattern,size,count,total,support")
f <<= nm.msortf(f="support%nr", o=self.pFile)
f.run()
if self.outtf:
# 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 ...")
self.tFile = self.temp.file()
xxw = tf.file() #Mtemp.new.name
f = None
f <<= nm.mcut(f=self.db.idFN, i=self.db.file)
f <<= nm.muniq(k=self.db.idFN)
f <<= nm.mnumber(S=0, a="__tid", q=True)
f <<= nm.msortf(f="__tid", o=xxw)
f.run()
translt = self.temp.file()
extTake.lcmtrans(lcmout, "t", translt)
f = None
f <<= nm.msortf(f="__tid", i=translt)
f <<= nm.mjoin(k="__tid", m=xxw, f=self.db.idFN)
f <<= nm.mcut(f=self.db.idFN + ",pid")
f <<= nm.msortf(f=self.db.idFN + ",pid", o=self.tFile)
f.run()
def run(self):
from datetime import datetime
t = datetime.now()
wf = nu.Mtemp()
xxinp = wf.file()
xxmap = wf.file()
input = self.ei
self.g2pair(self.ni, self.nf, self.ei, self.ef1, self.ef2, xxinp,
xxmap)
xxmace = wf.file() # maceの出力(tra形式)
para = {}
if self.msgoff:
para["type"] = "Ce_" if self.all else "Me_"
else:
para["type"] = "Ce" if self.all else "Me"
para["i"] = xxinp
para["o"] = xxmace
if self.minSize:
para["l"] = self.minSize
if self.maxSize:
para["u"] = self.maxSize
extTake.mace(para)
#MCMD::msgLog("converting the numbered nodes into original name ...")
id = nu.mrecount(i=xxmace, nfni=True)
# xxpair = wf.file() # 上記traをpair形式に変換したデータ
fpair = None
fpair <<= nm.mcut(i=xxmace, nfni=True, f="0:num")
fpair <<= nm.mnumber(q=True, a="id")
fpair <<= nm.mvcount(vf="num:size")
fpair <<= nm.mtra(r=True, f="num")
# when ni= specified, it add the isolated single cliques.
if self.ni:
fpair_u = nm.mread(i=fpair)
if self.all:
fpair_u <<= nm.mselstr(f="size", v=1)
fpair_u <<= nm.mcut(f="num")
fpair_u <<= nm.muniq(k="num")
# select all nodes which are not included in any cliques
xxiso = None
xxiso <<= nm.mcut(f="num", i=xxmap)
xxiso <<= nm.mcommon(k="num", m=fpair_u, r=True)
xxiso <<= nm.mnumber(S=id, a="id", q=True)
xxiso <<= nm.msetstr(v=1, a="size")
xxiso <<= nm.mcut(f="id,num,size")
# mcut入れないとおかしくなるあとで直す
#ddlist = [fpair.mcut(f="id,num,size"),xxiso]
xxpair = nm.mcut(i=[fpair, xxiso], f="id,num,size")
else:
xxpair = fpair
xxpair <<= nm.mjoin(m=xxmap, k="num", f="node")
xxpair <<= nm.mcut(f="id,node,size")
xxpair <<= nm.msortf(f="id,node", o=self.oFile)
xxpair.run()
procTime = datetime.now() - t
# ログファイル出力
if self.logFile:
kv = [["key", "value"]]
for k, v in self.args.items():
kv.append([k, str(v)])
kv.append(["time", str(procTime)])
nm.writecsv(i=kv, o=self.logFile).run()
def __mkNode(key, nf, nl, nv, nc, ni, ef1, ef2, ei, noiso, norm, mapFile,
oFile):
xbyE = None
xbyN = None
# edgeファイルからnode情報を生成
# noiso(孤立node排除)の場合は、edgeにあってnodeにないidを省く必要があるので計算する。
if ni == None or (ni != None and noiso):
inp = [
nm.mcut(f="%s:key,%s:nam,%s:nl" % (key, ef1, ef1), i=ei),
nm.mcut(f="%s:key,%s:nam,%s:nl" % (key, ef2, ef2), i=ei)
]
xbyE <<= nm.mnullto(i=inp, f="key", v="##NULL##")
xbyE <<= nm.muniq(k="key,nam")
xbyE <<= nm.mjoin(k="key", K="nam", m=mapFile, f="num:keyNum")
xbyE <<= nm.mjoin(k="nam", K="nam", m=mapFile, f="num,leaf")
xbyE <<= nm.msetstr(v=",,,,", a="nv,nc,nlKey,nvKey,ncKey")
xbyE <<= nm.mcut(
f="key,nam,keyNum,num,nl,nv,nc,leaf,nvKey,ncKey,nlKey")
# nodeファイルから作成
if ni:
# mcal cat用のlabel項目の作成
label = []
#label項目
if nl:
for nml in nl:
label.append(nml)
else:
label.append("$s{%s}" % (nf))
nvcStr = []
if nv:
nvcStr.append('%s:nv' % (nv))
if nc:
nvcStr.append('%s:nc' % (nc))
"""
# map
# nam,leaf,num
# ##NULL##,,0
# #1_1,,2
# #1_2,,3
# #1_3,,4
# #2_1,,5
# a,1,6
# b,1,7
# c,1,8
"""
f = None
f <<= nm.mcal(c='cat("_",%s)' % (','.join(label)), a="##label", i=ni)
if len(nvcStr) == 0:
f <<= nm.mcut(f='%s:key,%s:nam,##label:nl' % (key, nf))
else:
f <<= nm.mcut(f='%s:key,%s:nam,##label:nl,%s' %
(key, nf, ','.join(nvcStr)))
f <<= nm.mnullto(f="key", v="##NULL##")
if not nv:
f <<= nm.msetstr(v="", a="nv")
if not nc:
f <<= nm.msetstr(v="", a="nc")
f <<= nm.mjoin(k="key", K="nam", m=mapFile, f="num:keyNum")
f <<= nm.mjoin(k="nam", K="nam", m=mapFile, f="num,leaf")
if norm:
#ノードの拡大率は固定
nr = 3
f <<= nm.mnormalize(f="nv:nv2", c="range")
f <<= nm.mcal(c='${nv2}*(%s-1)+1' % (nr), a="nvv")
f <<= nm.mcut(f="key,nam,keyNum,num,nl,nvv:nv,nc,leaf") #o=#{xxa}"
else:
f <<= nm.mcut(f="key,nam,keyNum,num,nl,nv,nc,leaf") #o=#{xxa}"
xbyN <<= nm.mjoin(k="keyNum",
K="num",
m=f,
f="nl:nlk,nv:nvKey,nc:ncKey",
n=True,
i=f)
xbyN <<= nm.mcal(c='if(isnull($s{nlk}),$s{key},$s{nlk})', a='nlKey')
xbyN <<= nm.mcut(f="nlk", r=True)
if ni != None and noiso:
nm.mcommon(k="key,nam", m=xbyE, i=xbyN, o=oFile).run()
elif ni != None:
xbyN.writecsv(o=oFile).run()
else:
xbyE.writecsv(o=oFile).run()
def enumerate(self, eArgs):
pFiles = []
tFiles = []
tf = mtemp.Mtemp()
for cName, posSize in self.db.clsNameRecSize.items():
negSize = self.db.traSize - posSize
if "minGR" in eArgs:
self.minGR = eArgs["minGR"]
else:
minProb = eArgs["minProb"] if ("minProb" in eArgs) else 0.5
if "uniform" in eArgs and eArgs["uniform"] == True:
self.minGR = (minProb / (1 - minProb)) * (
self.db.clsSize - 1) # マニュアルの式(4)
else:
self.minGR = (minProb / (1 - minProb)) * (
float(negSize) / float(posSize)) # マニュアルの式(4)
# 最小サポートと最小サポート件数
# s=0.05
# s=c1:0.05,c2:0.06
# S=10
# S=c1:10,c2:15
if "minCnt" in eArgs:
if isinstance(eArgs["minCnt"], dict):
self.minPos = eArgs["minCnt"][cName]
else:
self.minPos = eArgs["minCnt"]
else:
if isinstance(eArgs["minSup"], dict):
self.minPos = int(eArgs["minSup"][cName] * float(posSize) +
0.99)
else:
self.minPos = int(eArgs["minSup"] * flost(posSize) + 0.99)
# 最大サポートと最大サポート件数
if "maxCnt" in eArgs:
if isinstance(eArgs["maxCnt"], dict):
self.maxPos = eArgs["maxCnt"][cName]
else:
self.maxPos = eArgs["maxCnt"]
elif "maxSup" in eArgs:
if isinstance(eArgs["maxSup"], dict):
self.maxPos = int(eArgs["maxSup"][cName] * float(posSize) +
0.99)
else:
self.maxPos = int(eArgs["maxSup"] * float(posSize) + 0.99)
else:
self.maxPos = None
self.sigma[cName] = self.calSigma(self.minPos, self.minGR, posSize,
negSize)
# lcmのパラメータ設定と実行
# 頻出パターンがなかった場合、lcm出力ファイルが生成されないので
# そのときのために空ファイルを生成しておいく。
lcmout = tf.file() # lcm出力ファイル
with open(lcmout, "w") as efile:
pass
runPara = {}
if self.msgoff:
runPara["type"] = eArgs["type"] + "IA_"
else:
runPara["type"] = eArgs["type"] + "IA"
#if self.maxPos: #rubyだとif @maxCntなってる（どこにも設定されてないので）動いてないはず
if self.maxPos:
runPara["U"] = self.maxPos
if "minLen" in eArgs:
runPara["l"] = str(eArgs["minLen"])
if "maxLen" in eArgs:
runPara["u"] = str(eArgs["maxLen"])
runPara["w"] = self.weightFile[cName]
runPara["i"] = self.file
runPara["sup"] = str(self.sigma[cName])
runPara["o"] = lcmout
# lcm実行
#MCMD::msgLog("#{run}")
#TAKE::run_lcm(run)
#print(self.sigma)
#print(runPara)
#MCMD::msgLog("output patterns to CSV file ...")
extTake.lcm(runPara)
pFiles.append(self.temp.file())
transle = tf.file()
extTake.lcmtrans(lcmout, "e", transle)
f = nm.mdelnull(f="pattern", i=transle)
f <<= nm.mcal(c='round(${countN},1)', a="neg")
f <<= nm.mcal(c='round(${countP}/%s,1)' % (self.posWeight[cName]),
a="pos")
f <<= nm.mdelnull(f="pattern") #いる？
f <<= nm.msetstr(v=cName, a="class")
f <<= nm.msetstr(v=posSize, a="posTotal")
f <<= nm.msetstr(v=self.minGR, a="minGR")
f <<= nm.mcut(f="class,pid,pattern,size,pos,neg,posTotal,minGR",
o=pFiles[-1])
f.run()
#s = nutil.mrecount(i=self.file)
#MCMD::msgLog("the number of contrast patterns on class `#{cName}' enumerated is #{s}")
if self.outtf:
# トランザクション毎に出現するパターンを書き出す
#MCMD::msgLog("output tid-patterns ...")
tFiles.append(self.temp.file())
xxw = tf.file()
xxw = nm.mcut(f=self.db.idFN, i=self.db.file)
xxw <<= nm.muniq(k=self.db.idFN)
xxw <<= nm.mnumber(S=0, a="__tid", q=True)
translt = self.temp.file()
extTake.lcmtrans(lcmout, "t", translt)
f = nm.mjoin(k="__tid", m=xxw, f=self.db.idFN, i=translt)
f <<= nm.msetstr(v=cName, a="class")
f <<= nm.mcut(f=self.db.idFN + ",class,pid", o=tFiles[-1])
f.run()
# クラス別のパターンとtid-pidファイルを統合して最終出力
self.pFile = self.temp.file()
self.tFile = self.temp.file()
# パターンファイル併合
xxpCat = tf.file()
f = nm.mcat(i=",".join(pFiles))
f <<= nm.msortf(f="class,pid")
f <<= nm.mnumber(s="class,pid", S=0, a="ppid", o=xxpCat)
f.run()
# パターンファイル計算
items = self.db.items
f = nm.mcut(f="class,ppid:pid,pattern,size,pos,neg,posTotal,minGR",
i=xxpCat)
f <<= nm.msetstr(v=self.db.traSize, a="total")
f <<= nm.mcal(c='${total}-${posTotal}', a="negTotal") # negのトータル件数
f <<= nm.mcal(c='${pos}/${posTotal}', a="support") # サポートの計算
f <<= nm.mcal(
c=
'if(${neg}==0,1.797693135e+308,(${pos}/${posTotal})/(${neg}/${negTotal}))',
a="growthRate")
if "uniform" in eArgs and eArgs["uniform"] == True:
f <<= nm.mcal(
c='(${pos}/${posTotal})/(${pos}/${posTotal}+(%s-1)*${neg}/${negTotal})'
% (self.db.clsSize),
a="postProb")
else:
f <<= nm.mcal(c='${pos}/(${pos}+${neg})', a="postProb")
f <<= nm.msel(c='${pos}>=%s&&${growthRate}>=${minGR}' %
(self.minPos)) # minSupとminGRによる選択
f <<= nm.mvreplace(vf="pattern",
m=items.file,
K=items.idFN,
f=items.itemFN)
f <<= nm.mcut(
f="class,pid,pattern,size,pos,neg,posTotal,negTotal,total,support,growthRate,postProb"
)
f <<= nm.mvsort(vf="pattern")
f <<= nm.msortf(f="class%nr,postProb%nr,pos%nr", o=self.pFile)
f.run()
# アイテムを包含している冗長なタクソノミを削除
if items.taxonomy:
#MCMD::msgLog("reducing redundant rules in terms of taxonomy ...")
##ここは後で
zdd = VSOP.constant(0)
dt = nm.mcut(i=self.pFile, f="pattern")
for fldVal in dt:
zdd = zdd + VSOP.itemset(fldVal[0])
zdd = self.reduceTaxo(zdd, self.db.items)
xxp1 = tf.file()
xxp2 = tf.file()
xxp3 = tf.file()
zdd.csvout(xxp1)
nm.mcut(nfni=True, f="1:pattern",
i=xxp1).mvsort(vf="pattern").msortf(f="pattern",
o=xxp2).run()
nm.msortf(f="pattern", i=self.pFile).mcommon(
k="pattern", m=xxp2).msortf(f="class%nr,postProb%nr,pos%nr",
o=xxp3).run()
shutil.move(xxp3, self.pFile)
if self.outtf:
# 列挙されたパターンを含むtraのみ選択するためのマスタ
xxp4 = nm.mcut(f="class,pid", i=self.pFile)
f = nm.mcat(i=",".join(tFiles))
f <<= nm.mjoin(k="class,pid", m=xxpCat,
f="ppid") # 全クラス統一pid(ppid)結合
f <<= nm.mcommon(k="class,ppid", K="class,pid",
m=xxp4) # 列挙されたパターンの選択
f <<= nm.mcut(f=self.db.idFN + ",class,ppid:pid")
f <<= nm.msortf(f=self.db.idFN + ",class,pid", o=self.tFile)
f.run()
def mgv(ei,
ef,
ev=None,
ec=None,
el=None,
ed=None,
ni=None,
nf=None,
nv=None,
nc=None,
nl=None,
nw=1,
tp="flat",
k=None,
o=None,
d=False,
clusterLabel=False,
noiso=False,
normalize=False,
normalizeEdge=False,
normalizeNode=False):
# arg check
# ei : str (filename)
# ef : str | list (fldname size=2)
# ev : str | None (fldname)
# ec : str | None (fldname)
# el : str | list | None (fldname no limit )
# ed : str | None (fldname)
# ni : str | None (filename)
# nf : str | None (fldname)
# nv : str | None (fldname)
# nc : str | None (fldname)
# nl : str | list | None (fldname no limit )
# tp : str (flat|nest default:flat )
# k : str | None (fldname)
# o : str (filename)
# d : bool | None
# clusterLabel : bool | None
# noiso : bool | None
# ei
if not (isinstance(ei, str)):
raise TypeError("ei= unsupport " + str(type(ei)))
# ef
if isinstance(ef, str):
ef = ef.split(',')
elif not isinstance(ef, list):
raise TypeError("ef= unsupport " + str(type(ef)))
if len(ef) < 2:
raise TypeError("ef size == 2 ")
elif len(ef) > 2:
sys.stderr.write('warning : ef size == 2 ')
# k
if not (isinstance(k, str) or k == None):
raise TypeError("k= unsupport " + str(type(k)))
# ev
if not (isinstance(ev, str) or ev == None):
raise TypeError("ev= unsupport " + str(type(ev)))
# ec
if not (isinstance(ec, str) or ec == None):
raise TypeError("ec= unsupport " + str(type(ec)))
# el
if isinstance(el, str):
el = el.split(',')
if len(el) == 1 and el[0] == '':
el = None
elif not (isinstance(el, list) or el == None):
raise TypeError("el= unsupport " + str(type(el)))
# ed
if not (isinstance(ed, str) or ed == None):
raise TypeError("ed= unsupport " + str(type(ed)))
# ni
if not (isinstance(ni, str) or ni == None):
raise TypeError("ni= unsupport " + str(type(ni)))
# nf
if not (isinstance(nf, str) or nf == None):
raise TypeError("nf= unsupport " + str(type(nf)))
# nv
if not (isinstance(nv, str) or nv == None):
raise TypeError("nv= unsupport " + str(type(nv)))
# nc
if not (isinstance(nc, str) or nc == None):
raise TypeError("nc= unsupport " + str(type(nc)))
# nl
if isinstance(nl, str):
nl = nl.split(',')
if len(nl) == 1 and nl[0] == '':
nl = None
elif not (isinstance(nl, list) or nl == None):
raise TypeError("nl= unsupport " + str(type(nl)))
# tp
if tp == None:
tp = "flat"
elif not isinstance(tp, str):
raise TypeError("tp= unsupport " + str(type(tp)))
# o
if isinstance(o, str):
oFile = o
else:
raise TypeError("o= unsupport " + str(type(o)))
# d
if d == None:
d = False
if not isinstance(d, bool):
raise TypeError("d= unsupport " + str(type(d)))
# clusterLabel
if clusterLabel == None:
clusterLabel = False
if not isinstance(clusterLabel, bool):
raise TypeError("clusterLabel= unsupport " + str(type(clusterLabel)))
# noiso
if noiso == None:
noiso = False
if not isinstance(noiso, bool):
raise TypeError("noiso= unsupport " + str(type(noiso)))
# noiso
if normalize == None:
normalize = False
if not isinstance(normalize, bool):
raise TypeError("noiso= unsupport " + str(type(normalize)))
if normalizeEdge == None:
normalizeEdge = False
if not isinstance(normalizeEdge, bool):
raise TypeError("noiso= unsupport " + str(type(normalizeEdge)))
if normalizeNode == None:
normalizeNode = False
if not isinstance(normalizeNode, bool):
raise TypeError("noiso= unsupport " + str(type(normalizeNode)))
if normalize:
normalizeEdge = True
normalizeNode = True
temp = mtemp.Mtemp()
xxni = temp.file()
xxei = temp.file()
xxmap = temp.file()
xxnode = temp.file()
xxedge = temp.file()
xxdotNode = temp.file()
xxdotEdge = temp.file()
xxtree = temp.file()
mkDir(xxdotNode)
mkDir(xxdotEdge)
if d:
directedStr = "edge []"
else:
directedStr = "edge [dir=none]"
# key追加 (cluster用)
if not k:
if ni:
nm.msetstr(v="", a="#key", i=ni, o=xxni).run()
ni = xxni
nm.msetstr(v="", a="#key", i=ei, o=xxei).run()
ei = xxei
k = "#key"
#efs = ef.split(",")
ef1 = ef[0]
ef2 = ef[1]
__mkMap(k, nf, ni, ef1, ef2, ei, xxmap)
__mkNode(k, nf, nl, nv, nc, ni, ef1, ef2, ei, noiso, normalizeNode, xxmap,
xxnode)
__mkEdge(k, ef1, ef2, el, ec, ed, ev, ei, normalizeEdge, xxmap, xxedge)
# dot用のnodeとedgeデータをcluster別ファイルとして生成
__dotNode(xxnode, nw, tp, clusterLabel, xxdotNode)
__dotEdge(xxedge, xxdotEdge)
depth = None
if tp == "flat":
depth = __mkFlat(xxnode, xxtree) # mgvとおなじ
elif tp == "nest":
# tree構造の処理
# クラスタのみtree構造に格納する
depth = __mkTree(xxnode, xxtree) # mgvとおなじ
else:
raise TypeError("unsupport type " + tp)
xxdotTree = temp.file()
header = '''digraph G {{
{directedStr}
'''.format(directedStr=directedStr)
footer = "}\n"
__dotTree(xxtree, depth, header, footer, xxdotTree) # mgvとおなじ
__replace(xxdotTree, xxdotNode, xxdotEdge, clusterLabel, o)
def run(self, **kw_args):
os.environ['KG_ScpVerboseLevel'] = "2"
if "msg" in kw_args:
if kw_args["msg"] == "on":
os.environ['KG_ScpVerboseLevel'] = "4"
ln = "#{@pt}line"
# make the line number
ln = "{}line".format(self.pt)
xxmap = self.workf.file()
sdata = self.workf.file()
# convert the data for sketchport
# mkdata
xx1 = nm.mnumber(S=0, a=ln, q=True, i=self.iFile)
if self.wfH:
xx2 = nm.mcut(f=self.wfH + self.tidH + self.elem, i=xx1)
else:
self.wfH = ["{}wf".format(self.pt)]
xx2 = nm.msetstr(v=0, a=self.wfH, i=xx1)
xx2 <<= nm.mcut(f=self.wfH + self.tidH + self.elem)
fmap = nm.mcut(f=[ln] + self.tidH, i=xx1, o=xxmap)
xx2 <<= nm.mcut(f=self.wfH + self.elem, nfno=True)
xx2 <<= nm.cmd("tr ',' ' '")
xx2 <<= nm.mwrite(o=sdata)
nm.runs([fmap, xx2])
# do sort
outf = self.workf.file()
para = {}
if self.dist == "C":
para["cosdist"] = self.th
elif self.dist == "H":
para["hamdist"] = self.th
if not self.uc:
para["centering"] = True
para["auto"] = True
para["windowsize"] = self.ws
para["seed"] = self.seed
para["missingratio"] = self.mr
para["i"] = sdata
para["o"] = outf
status = extMining.sketchsort(para)
if status:
raise Exception("#ERROR# checking sketchsort messages")
tmp = []
for val in self.tidH:
tmp.append("{}:{}2".format(val, val))
tid2 = ",".join(tmp)
f = nm.mread(i=outf)
f <<= nm.cmd("tr ' ' ',' ")
f <<= nm.mcut(nfni=True, f="0:eline1,1:eline2,2:distance")
f <<= nm.mfsort(f="eline*")
# 行番号に対応するtidを取得
f <<= nm.mjoin(k="eline1",
K="{}line".format(self.pt),
f=self.tidH,
m=xxmap)
f <<= nm.mjoin(k="eline2", K="{}line".format(self.pt), f=tid2, m=xxmap)
f <<= nm.msortf(f="eline1%n,eline2%n")
f <<= nm.mcut(r=True, f="eline1,eline2")
f <<= nm.msortf(f=self.tidH)
f <<= nm.mfldname(q=True, o=self.oFile)
f.run()
nu.mmsg.endLog(self.__cmdline())
def enumerate(self,eArgs):
"""
eArgsで与えられた条件で、頻出アイテム集合の列挙を実行する。
:type eArgs: dict
:type eArgs['type']: str
:type eArgs['minCnt']: int
:type eArgs['minSup']: float
:type eArgs['maxCnt']: int
:type eArgs['maxSup']: float
:type eArgs['minLen']: int
:type eArgs['maxLen']: int
:type eArgs['top']: int
:type eArgs['skipTP']: bool【default:False】
:param eArgs: 各種列挙パラメータ
:param eArgs['type']: 抽出するアイテム集合の型【'F':頻出集合, 'C':飽和集合, 'M':極大集合】
:param eArgs['minCnt']: 最小サポート(件数)
:param eArgs['minSup']: 最小サポート(確率)
:param eArgs['maxCnt']: 最大サポート(件数)
:param eArgs['maxSup']: 最大サポート(確率)
:param eArgs['minLen']: アイテム集合の最小アイテム数(件数)
:param eArgs['maxLen']: アイテム集合の最大アイテム数(件数)
:param eArgs['top']: 列挙するサポート上位件数(件数)
:param eArgs['skipTP']: トランザクションにマッチするパターン(アイテム集合)の出力を行わない。
"""
tf=mtemp.Mtemp()
self.eArgs=eArgs
self.type = eArgs["type"]
if "minCnt" in eArgs and eArgs["minCnt"] != None:
self.minCnt = int(eArgs["minCnt"])
self.minSup = float(self.minCnt) / float(self.db.traSize)
else:
self.minSup = float(eArgs["minSup"])
self.minCnt = int(self.minSup * float(self.db.traSize) + 0.99)
# 最大サポートと最大サポート件数
self.maxCnt=None
if ("maxCnt" in eArgs and eArgs["maxCnt"]!= None) or ( "maxSup" in eArgs and eArgs["maxSup"]!= None):
if "maxCnt" in eArgs and eArgs["maxCnt"]!= None:
self.maxCnt = int(eArgs["maxCnt"])
self.maxSup = float(self.maxCnt) / float(self.db.traSize)
else:
self.maxSup = float(eArgs["maxSup"])
self.maxCnt = int(self.maxSup * float(self.db.traSize) + 0.99)
params = {}
if self.msgoff:
params["type"] ="%sIf_"%(self.type)
else:
params["type"] ="%sIf"%(self.type)
if self.maxCnt :
params["U"] = str(self.maxCnt)
if "minLen" in eArgs and eArgs["minLen"] != None :
params["l"] = str(eArgs['minLen'])
if "maxLen" in eArgs and eArgs["maxLen"] != None :
params["u"] = str(eArgs['maxLen'])
# 列挙パターン数上限が指定されれば、一度lcmを実行して最小サポートを得る
if "top" in eArgs and eArgs["top"] != None :
self.top = eArgs["top"]
if self.top and self.top>0 :
xxtop = tf.file()
import copy
top_params = copy.deepcopy(params)
top_params["i"] = self.file
top_params["sup"] = "1"
top_params["K"] = str(self.top)
top_params["so"] = xxtop
import re
top_params["type"] = re.sub('_$', '', top_params["type"] )
extTake.lcm(top_params)
with open(xxtop, "r") as rfile:
self.minCnt = int(rfile.read().strip())
if self.minCnt<0 :
self.minCnt=1
self.skipTP=False
if "skipTP" in eArgs:
self.skipTP=eArgs["skipTP"]
# lcm_seq出力ファイル
lcmout = tf.file()
# 頻出パターンがなかった場合、lcm出力ファイルが生成されないので
# そのときのために空ファイルを生成しておいく。
with open(lcmout, "w") as efile:
pass
# lcm実行
params["i"] = self.file
params["sup"] = str(self.minCnt)
params["o"] = lcmout
extTake.lcm(params)
# caliculate one itemset for lift value
xxone= tf.file()
tpstr = "FIf_" if self.msgoff else "FIf"
extTake.lcm(type=tpstr,i=self.file,sup=1,o=xxone,l=1,u=1)
# パターンのサポートを計算しCSV出力する
#MCMD::msgLog("output patterns to CSV file ...")
xxp0 = tf.file()
self.pFile = self.temp.file()
items=self.db.items
trans0 = self.temp.file()
extTake.lcmtrans(lcmout,"p",trans0)
f = nm.mdelnull(i=trans0,f="pattern")
f <<= nm.mvreplace(vf="pattern",m=items.file,K=items.idFN,f=items.itemFN)
f <<= nm.msetstr(v=self.db.traSize,a="total")
f <<= nm.mcal(c='${count}/${total}',a="support")
f <<= nm.mcut(f="pid,pattern,size,count,total,support")
f <<= nm.mvsort(vf="pattern")
f <<= nm.msortf(f="pid",o=xxp0)
f.run()
# xxp0
# pid,count,total,support,pattern
# 0,13,13,1,A
# 4,6,13,0.4615384615,A B
xxp1=tf.file()
# taxonomy指定がない場合(2010/11/20追加)
if items.taxonomy==None:
shutil.move(xxp0,xxp1)
# taxonomy指定がある場合
else:
#MCMD::msgLog("reducing redundant rules in terms of taxonomy ...")
zdd=VSOP.constant(0)
fobj = nm.mcut(i=xxp0,f='pattern')
for fldVal in fobj:
zdd=zdd+VSOP.itemset(fldVal[0])
zdd=self.reduceTaxo(zdd,self.db.items)
xxz1=tf.file()
xxz2=tf.file()
zdd.csvout(xxz1)
f0=None
f0 <<= nm.mcut(nfni=True,f="1:pattern",i=xxz1)
f0 <<= nm.mvsort(vf="pattern")
f0 <<= nm.msortf(f="pattern")
f=None
f <<= nm.msortf(f="pattern",i=xxp0)
f <<= nm.mcommon(k="pattern",m=f0)
f <<= nm.msortf(f="pid",o=xxp1)
f.run()
# lift値の計算
transl = tf.file()
extTake.lcmtrans(xxone,"p",transl)
xxp2 = nm.mdelnull(i=transl,f="pattern")
xxp2 <<= nm.mvreplace(vf="pattern",m=items.file,K=items.idFN,f=items.itemFN)
xxp2 <<= nm.msortf(f="pattern")
xxp3 = nm.mcut(f="pid,pattern",i=xxp1)
xxp3 <<= nm.mtra(f="pattern",r=True)
xxp3 <<= nm.mjoin(k="pattern",m=xxp2,f="count:c1")
xxp3 <<= nm.mcal(c='ln(${c1})',a="c1ln")
xxp3 <<= nm.msum(k="pid",f="c1ln")
# p3
# pid,pattern,c1,c1ln
# 0,A,13,2.564949357
# 1,E,7,1.945910149
#おかしくなる?=>OK
f3 = nm.mjoin(k="pid",f="c1ln",i=xxp1,m=xxp3)
f3 <<= nm.mcal(c='round(exp(ln(${count})-${c1ln}+(${size}-1)*ln(${total})),0.0001)',a="lift")
f3 <<= nm.mcut(f="pid,size,count,total,support,lift,pattern")
f3 <<= nm.msortf(f="support%nr",o=self.pFile)
f3.run()
#self.size = mrecount.mrecount(i=self.file)
#MCMD::msgLog("the number of patterns enumerated is #{@size}")
if not self.skipTP:
# 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 ...")
self.tFile = self.temp.file()
xxw3i = tf.file()
extTake.lcmtrans(lcmout,"t",xxw3i)
xxw1 = nm.mcut(f=self.db.idFN,i=self.db.file).muniq(k=self.db.idFN).mnumber(S=0,a="__tid",q=True).msortf(f="__tid")
xxw2 = nm.mcut(f="pid",i=self.pFile)
xxw3 = nm.mcommon(k="pid",i=xxw3i,m=xxw2).mjoin(k="__tid",m=xxw1,f=self.db.idFN).mcut(f=self.db.idFN+",pid",o=self.tFile)
xxw3.run()
def enumerate(self,eArgs):
tf=nu.Mtemp()
# 最小サポートと最小サポート件数
if "minCnt" in eArgs :
self.minCnt = int(eArgs["minCnt"])
self.minSup = float(self.minCnt)/ float(self.db.size)
else:
self.minSup = float(eArgs["minSup"])
self.minCnt = int(self.minSup * float(self.db.size) + 0.99)
# 最大サポートと最大サポート件数
self.maxCnt=None
if "maxCnt" in eArgs or "maxSup" in eArgs:
if "maxCnt" in eArgs:
self.maxCnt = int(eArgs["maxCnt"])
self.maxSup = float(self.maxCnt)/float(self.db.size)
else:
self.maxSup = float(eArgs["maxSup"])
self.maxCnt = int(self.maxSup * float(self.db.size) + 0.99)
#未使用
#@minProb = eArgs["minProb"].to_f # 事後確率
#@minGR = @minProb/(1-@minProb) # 増加率
#@minGR = eArgs["minGR"].to_f if eArgs["minGR"]
# あるクラスをpos、他のクラスをnegにして、パターン列挙した結果ファイル名を格納する
pFiles=[]
tFiles=[]
for cName,posSize in self.db.clsNameRecSize.items():
negSize=self.db.size-posSize
# minGRの計算
if "minGR" in eArgs:
self.minGR=eArgs["minGR"]
else:
minProb = eArgs["minProb"] if ( "minProb" in eArgs ) else 0.5
if "uniform" in eArgs and eArgs["uniform"]:
self.minGR = (minProb/(1-minProb)) * (self.db.clsSize-1) # マニュアルの式(4)
else:
self.minGR = (minProb/(1-minProb)) * (float(negSize)/float(posSize)) # マニュアルの式(4)
# 最小サポートと最小サポート件数
if "minCnt" in eArgs:
self.minPos = eArgs["minCnt"]
else:
self.minPos = int(eArgs["minSup"] * float(posSize) + 0.99)
# 最大サポートと最大サポート件数
if "maxCnt" in eArgs or "maxSup" in eArgs:
if "maxCnt" in eArgs:
self.maxCnt = int(eArgs["maxCnt"])
else:
self.maxCnt = int(eArgs["maxSup"] * float(posSize) + 0.99)
self.sigma[cName] = self.calSigma(self.minPos,self.minGR,posSize,negSize)
# lcm_seqのパラメータ設定と実行
lcmout = tf.file() # lcm_seq出力ファイル
# 頻出パターンがなかった場合、lcm出力ファイルが生成されないので
# そのときのために空ファイルを生成しておいく。
with open(lcmout, "w") as efile:
pass
params = {}
if self.msgoff:
params["type"] ="CIA_"
else:
params["type"] ="CIA"
if self.maxCnt: # windowサイズ上限
params["U"] = str(self.maxCnt)
if "minLen" in eArgs:
params["l"] = str(eArgs["minLen"])
if 'maxLen' in eArgs:
params["u"] = str(eArgs["maxLen"])
if 'gap' in eArgs:
params["g"] = str(eArgs["gap"])
if 'win' in eArgs:
params["G"] = str(eArgs["win"])
params["w"] = self.weightFile[cName]
params["i"] = self.file
params["sup"] = str(self.sigma[cName])
params["o"] = lcmout
# lcm_seq実行
#MCMD::msgLog("#{run}")
if 'padding' in eArgs and eArgs["padding"]: # padding指定時は、0アイテムを出力しないlcm_seqを実行
extTake.lcmseq_zero(params)
else:
extTake.lcmseq(params)
# 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 ...")
pFiles.append(self.temp.file())
transle = self.temp.file()
extTake.lcmtrans(lcmout,"e",transle) # pattern,countP,countN,size,pid
f=None
f <<= nm.mdelnull(f="pattern",i=transle)
f <<= nm.mcal(c='round(${countN},1)',a="neg")
f <<= nm.mcal(c='round(${countP}/%s,1)'%(self.posWeight[cName]),a="pos")
f <<= nm.mdelnull(f="pattern")
f <<= nm.msetstr(v=cName,a="class")
f <<= nm.msetstr(v=posSize,a="posTotal")
f <<= nm.msetstr(v=self.minGR,a="minGR")
f <<= nm.mcut(f="class,pid,pattern,size,pos,neg,posTotal,minGR",o=pFiles[-1])
f.run()
#s = MCMD::mrecount("i=#{pFiles.last}") # 列挙されたパターンの数
#MCMD::msgLog("the number of contrast patterns on class `#{cName}' enumerated is #{s}")
if self.outtf :
# 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 ...")
tFiles.append(self.temp.file())
xxw= tf.file()
f=None
f <<= nm.mcut(f=self.db.idFN,i=self.db.file)
f <<= nm.muniq(k=self.db.idFN)
f <<= nm.mnumber(S=0,a="__tid",q=True)
f <<= nm.msortf(f="__tid",o=xxw)
f.run()
nm.mcut(f=self.db.idFN,i=self.db.file).muniq(k=self.db.idFN).mnumber(S=0,a="__tid",q=True,o=xxw).run()
translt = self.temp.file()
extTake.lcmtrans(lcmout,"t",translt)
nm.mjoin(k="__tid",m=xxw,f=self.db.idFN,i=translt).msetstr(v=cName,a="class").mcut(f=self.db.idFN+",class,pid",o=tFiles[-1]).run()
# クラス別のパターンとtid-pidファイルを統合して最終出力
self.pFile = self.temp.file()
self.tFile = self.temp.file()
# パターンファイル併合
xxpCat = tf.file()
f = nm.mcat(i=",".join(pFiles))
f <<= nm.msortf(f="class,pid")
f <<= nm.mnumber(s="class,pid",S=0,a="ppid",o=xxpCat)
f.run()
# パターンファイル計算
items=self.db.items
f=""
f = nm.mcut(f="class,ppid:pid,pattern,size,pos,neg,posTotal,minGR",i=xxpCat)
f <<= nm.msetstr(v=self.db.size,a="total")
f <<= nm.mcal(c='${total}-${posTotal}',a="negTotal") # negのトータル件数
f <<= nm.mcal(c='${pos}/${posTotal}',a="support") # サポートの計算
f <<= nm.mcal(c='if(${neg}==0,1.797693135e+308,(${pos}/${posTotal})/(${neg}/${negTotal}))',a="growthRate")
if "uniform" in eArgs and eArgs["uniform"] == True:
f <<= nm.mcal(c='(${pos}/${posTotal})/(${pos}/${posTotal}+(%s-1)*${neg}/${negTotal})'%(self.db.clsSize),a="postProb")
else:
f <<= nm.mcal(c='${pos}/(${pos}+${neg})',a="postProb")
f <<= nm.msel(c='${pos}>=%s&&${growthRate}>=${minGR}'%(self.minPos)) # minSupとminGRによる選択
f <<= nm.mvreplace(vf="pattern",m=items.file,K=items.idFN,f=items.itemFN)
f <<= nm.mcut(f="class,pid,pattern,size,pos,neg,posTotal,negTotal,total,support,growthRate,postProb")
f <<= nm.mvsort(vf="pattern")
f <<= nm.msortf(f="class%nr,postProb%nr,pos%nr",o=self.pFile)
f.run()
if self.outtf :
# 列挙されたパターンを含むtraのみ選択するためのマスタ
xxp4=nm.mcut(f="class,pid",i=self.pFile)
f = nm.mcat(i=",".join(tFiles))
f <<= nm.mjoin(k="class,pid",m=xxpCat,f="ppid") # 全クラス統一pid(ppid)結合
f <<= nm.mcommon(k="class,ppid",K="class,pid",m=xxp4) # 列挙されたパターンの選択
f <<= nm.mcut(f=self.db.idFN+",class,ppid:pid")
f <<= nm.msortf(f=self.db.idFN+",class,pid",o=self.tFile)
f.run()
self.size = nu.mrecount(i=self.pFile)
def mnetpie(ei,
ni,
ef,
nf,
o,
nodeSizeFld=None,
nodeTipsFld=None,
nodeColorFld=None,
edgeWidthFld=None,
edgeColorFld=None,
pieDataFld=None,
pieTipsFld=None,
picFld=None,
undirect=False,
offline=False):
#ei:edge file
#ef:egfile
if type(ef) is str:
ef = ef.split(',')
if len(ef) != 2:
raise Exception("ef= takes just two field names")
if not ((pieDataFld == None and pieTipsFld == None) or
(pieDataFld != None and pieTipsFld != None)):
raise Exception(
"pieDataFld= pieTipsFld= are necessary at the same time")
if picFld != None and pieDataFld != None:
raise Exception(
"picFld= cannot be specified with pieDataFld= pieTipsFld=")
if nodeColorFld != None:
if picFld != None or pieDataFld != None or pieTipsFld != None:
raise Exception(
"nodeColorFld= cannot be specified with pieDataFld= pieTipsFld= picFld="
)
if pieDataFld != None and pieTipsFld != None:
caseNo = 1
elif picFld != None:
caseNo = 2
else:
caseNo = 0
tempW = mtemp.Mtemp()
xxnode = tempW.file()
nodefld = []
nodedmy1 = []
nodedmy2 = []
nodefld.append("%s:node" % (nf))
if nodeSizeFld != None:
nodefld.append("%s:nodesize" % (nodeSizeFld))
else:
nodedmy1.append("nodesize")
nodedmy2.append("50")
if nodeTipsFld != None:
nodefld.append("%s:nodeT" % (nodeTipFld))
else:
nodedmy1.append("nodeT")
nodedmy2.append("")
if nodeColorFld != None:
nodefld.append("%s:nodeClr" % (nodeColorFld))
else:
nodedmy1.append("nodeClr")
nodedmy2.append("skyblue")
if caseNo == 1:
nodefld.append("%s:pieD" % (pieDataFld))
nodefld.append("%s:pieT" % (pieTipsFld))
elif caseNo == 2:
nodefld.append("%s:pic" % (picFld))
else:
nodedmy1.append("pic")
nodedmy2.append("")
f1 = None
f1 <<= nm.mcut(i=ni, f=nodefld)
if len(nodedmy1) != 0:
f1 <<= nm.msetstr(a=nodedmy1, v=nodedmy2)
if caseNo == 1:
f1 <<= nm.mshare(k="node", f="pieD:pieDS")
f1 <<= nm.mnumber(k="node", a="nodeid", B=True)
f2 = nm.muniq(k="pieT", i=f1)
f2 <<= nm.mnumber(q=True, a="pieTno")
f2 <<= nm.mjoin(k="pieT", f="pieTno", i=f1).iredirect("m")
f2 <<= nm.msortf(f="nodeid%n,pieTno%n", o=xxnode)
else:
f2 = nm.mnumber(a="nodeid%n", q=True, i=f1, o=xxnode)
f2.run()
xxedge = tempW.file()
# MAKE EDGE DATA
edgefld = []
edgedmy1 = []
edgedmy2 = []
edgefld.append("%s:edgeS" % (ef[0]))
edgefld.append("%s:edgeE" % (ef[1]))
if edgeWidthFld != None:
edgefld.append("%s:edgesize" % (edgeWidthFld))
else:
edgedmy1.append("edgesize")
edgedmy2.append("1")
if edgeColorFld != None:
edgefld.append("%s:edgecolor" % (edgeColorFld))
else:
edgedmy1.append("edgecolor")
edgedmy2.append("black")
f3 = None
f3 <<= nm.mcut(i=ei, f=edgefld)
if len(edgedmy1) != 0:
f3 <<= nm.msetstr(a=edgedmy1, v=edgedmy2)
f3 <<= nm.mnumber(a="preNo", q=True)
f3 <<= nm.mbest(k="edgeS,edgeE", s="preNo%nr")
f3 <<= nm.mnumber(s="preNo%n", a="edgeID")
f3 <<= nm.mjoin(k="edgeS", K="node", f="nodeid:edgeSid", m=xxnode)
f3 <<= nm.mjoin(k="edgeE", K="node", f="nodeid:edgeEid", m=xxnode)
#双方向チェック一応
f4 = None
f4 <<= nm.mfsort(i=f3, f="edgeS,edgeE")
f4 <<= nm.mcount(k="edgeS,edgeE", a="edgecnt")
f4 <<= nm.mselnum(c="[2,]", f="edgecnt")
f4 <<= nm.msetstr(a="biflg", v=1)
f4 <<= nm.mjoin(k="edgeID", f="biflg", n=True, i=f3).iredirect("m")
f4 <<= nm.msortf(f="edgeID%n", o=xxedge)
f4.run()
gdata = "{\"nodes\":["
if caseNo == 1:
nodedatastk = []
nodedatas = ""
for val, top, bot in nm.readcsv(xxnode).getline(k="nodeid",
otype='dict',
q=True):
name = val["node"]
r = val["nodesize"]
title = val["nodeT"]
if top:
nodedatas = "{\"name\":\"%s\",\"title\":\"%s\",\"r\":%s,\"node\":[" % (
name, title, r)
pieTno = val["pieTno"]
pieT = val["pieT"]
pieDS = val["pieDS"]
nodedatas += "{\"group\":%s,\"color\":%s,\"value\":%s,\"title\":\"%s\"}" % (
pieTno, pieDS, pieDS, pieT)
if bot:
nodedatas += "]}"
nodedatastk.append(nodedatas)
nodedatas = ""
else:
nodedatas += ","
gdata += ",".join(nodedatastk)
else:
nodedatastk = []
for val in nm.readcsv(xxnode).getline(otype='dict'):
name = val["node"]
r = val["nodesize"]
title = val["nodeT"]
pic = val["pic"]
nclr = val["nodeClr"]
nodedatas = "{\"name\":\"%s\",\"title\":\"%s\",\"pic\":\"%s\",\"color\":\"%s\",\"r\":%s}" % (
name, title, pic, nclr, r)
nodedatastk.append(nodedatas)
gdata += ",".join(nodedatastk)
gdata += "],\"links\": ["
edgedatastk = []
for val in nm.readcsv(xxedge).getline(otype='dict'):
es = val["edgeSid"]
et = val["edgeEid"]
esize = val["edgesize"]
ecolor = val["edgecolor"]
edgedatas = "{\"source\":%s,\"target\":%s,\"length\":500,\"ewidth\":%s,\"color\":\"%s\"}" % (
es, et, esize, ecolor)
edgedatastk.append(edgedatas)
gdata += ','.join(edgedatastk)
gdata += "]}"
direct = ".attr('marker-end','url(#arrowhead)')"
if undirect:
direct = ""
nodeTemplate = '''
node
.append("circle")
.attr("r",function(d){return d.r/4;})
.attr("fill", function(d){return d.color;})
.append("title")
.text(function(d){return d.title;})
'''
nodemakeTemplate = '''
for(var i=0 ; i< graph.nodes.length;i++){
graph.nodes[i].id = i
}
'''
if pieDataFld != None:
nodeTemplate = '''
node.selectAll("path")
.data( function(d, i){
return pie(d.node);
})
.enter()
.append("svg:path")
.attr("d", arc)
.attr("fill", function(d, i) {
return color(d.data.group);
})
.append("title")
.text(function(d){{return d.data.title;}})
node.append("circle")
.attr("r",function(d){{return d.r/4;}})
.attr({
'fill': 'white'
})
.append("title")
.text(function(d){{return d.title;}});
'''
nodemakeTemplate = '''
for(var i=0 ; i< graph.nodes.length;i++){
var r = graph.nodes[i].r
for(var j=0 ; j< graph.nodes[i].node.length;j++){
graph.nodes[i].node[j]['r'] = r
}
graph.nodes[i].id = i
}
'''
elif picFld != None:
nodeTemplate = '''
node
.append("image")
.attr("height",function(d){return d.r;})
.attr("width",function(d){return d.r;})
.attr("x",function(d){return -1 * d.r/2; })
.attr("y",function(d){return -1 * d.r/2; })
.attr("xlink:href",function(d){return d.pic; })
.append("title")
.text(function(d){return d.title;})
'''
d3js_str = "<script type='text/javascript' src='http://d3js.org/d3.v3.min.js'></script>"
if offline:
d3js_str = "<script>%s<script>" % (vjs.ViewJs.d3jsMin())
outTemplate = '''
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
{d3js_str}
<style></style>
</head>
<body>
<script type="text/javascript">
var graph = {gdata} ;
var width = 4000,
height = 3000;
var color = d3.scale.category10();
{nodemakeTemplate};
for(var i=0 ; i< graph.links.length;i++){{
graph.links[i].id = i
}}
var pie = d3.layout.pie()
.sort(null)
.value(function(d) {{ return d.value; }});
var arc = d3.svg.arc()
.outerRadius( function(d){{ return d.data.r ; }})
.innerRadius( function(d){{ return d.data.r/2 ; }} );
var svg = d3.select("body").append("svg")
.attr("width", width)
.attr("height", height);
d3.select("svg").append('defs').append('marker')
.attr({{'id':'arrowhead',
'viewBox':'-0 -5 10 10',
'refX':30,
'refY':0,
'orient':'auto-start-reverse',
'markerWidth':5,
'markerHeight':5,
'xoverflow':'visible'}})
.append('path')
.attr('d', 'M 0,-5 L 10 ,0 L 0,5')
.attr('fill', '#999')
.style('stroke','none');
var g = svg.append("g");
var node = g.selectAll(".node");
var link = g.selectAll(".link");
nodes = graph.nodes
links = graph.links
var force =
d3.layout.force()
.linkDistance(200)
.linkStrength(3.5)
.charge(-3500)
.gravity(0.1)
.friction(0.95)
.size([width, height])
.on("tick", function() {{
link
.attr("x1", function(d) {{ return d.source.x; }})
.attr("y1", function(d) {{ return d.source.y; }})
.attr("x2", function(d) {{ return d.target.x; }})
.attr("y2", function(d) {{ return d.target.y; }});
node
.attr("x", function(d) {{ return d.x; }})
.attr("y", function(d) {{ return d.y; }})
.attr("transform", function(d) {{ return "translate(" + d.x + "," + d.y + ")"}});
}});
node = node.data(nodes, function( d ) {{ return d.id; }} );
link = link.data(links, function( d ) {{ return d.id; }} );
link
.enter()
.append("line")
.attr("class", "link")
.style("stroke", function( d ) {{ return d.color; }} )
.style("stroke-width", function( d ) {{ return d.ewidth; }})
{direct}
node
.enter()
.append("g")
.attr("class", "node")
.style({{}})
.call(force.drag)
.on("contextmenu", function(nd) {{
d3.event.preventDefault();
force.stop()
nodes.splice( nd.index, 1 );
links = links.filter(function(nl) {{
return nl.source.index != nd.index && nl.target.index != nd.index;
}});
node = node.data(nodes, function( d ) {{ return d.id; }} );
node.exit().remove();
link = link.data( links, function( d ) {{ return d.id; }} );
link.exit().remove();
force.nodes(nodes)
.links(links)
.start();
}});
{nodeTemplate}
node
.append("text")
.attr("text-anchor", "middle")
.style("stroke", "black")
.text(function(d) {{ return d.name; }});
force.nodes(nodes)
.links(links)
.start();
</script>
</body>
</html>
'''.format(d3js_str=d3js_str,
gdata=gdata,
nodemakeTemplate=nodemakeTemplate,
direct=direct,
nodeTemplate=nodeTemplate)
html = sys.stdout
if not o == None:
html = open(o, "w")
html.write(outTemplate)
if not o == None:
html.close()
def run(self):
from datetime import datetime
t = datetime.now()
temp=nu.Mtemp()
xxsspcin=temp.file()
xxmap=temp.file()
# traファイルの変換とマップファイルの作成
if self.num :
total = self.convN(self.iFile,self.idFN,self.itemFN,xxsspcin,xxmap)
else:
total = self.conv(self.iFile,self.idFN,self.itemFN,xxsspcin,xxmap)
# system "head xxsspcin"
# 3 5 0 2
# 4 1 2
# 0 2 3 1
# 1 0 2
# 3 4 0 1
# system "head xxmap"
# ##item,##freq%0nr,##num
# b,4,0
# d,4,1
# f,4,2
minSupp = int(total*self.minSupPrb) if self.minSupPrb else self.minSupCnt
# sspc用simの文字列
if self.sim :
if self.sim=="J":
sspcSim="R"
elif self.sim=="P":
sspcSim="P"
elif self.sim=="C":
sspcSim="i"
# sim=省略時はRでth=0とする(sim制約なし)
else:
sspcSim="R"
self.th=0
############ 列挙本体 ############
xxsspcout=temp.file()
tpstr = sspcSim+"ft_" if self.msgoff else sspcSim+"ft"
extTake.sspc(type=tpstr,TT=minSupp,i=xxsspcin,th=self.th,o=xxsspcout)
##################################
xxtmmp=temp.file()
f = nm.mread(i=xxsspcout)
f <<= nm.cmd("tr ' ()' ','")
f <<= nm.mcut(f="1:i1,2:i2,0:frequency,4:sim",nfni=True)
if self.num :
f <<= nm.mfldname(f="i1:node1,i2:node2")
if self.sim!="C":
f <<= nm.mfsort(f="node1,node2")
f <<= nm.mjoin(k="node1",K="##item",m=xxmap,f="##freq:frequency1")
f <<= nm.mjoin(k="node2",K="##item",m=xxmap,f="##freq:frequency2")
else:
f <<= nm.mjoin(k="i1",K="##num",m=xxmap,f="##item:node1,##freq:frequency1")
f <<= nm.mjoin(k="i2",K="##num",m=xxmap,f="##item:node2,##freq:frequency2")
if self.sim!="C":
f <<= nm.mcut(f="i1,i2,frequency,sim,node1,node2,frequency1,frequency2,node1:node1x,node2:node2x")
f <<= nm.mfsort(f="node1x,node2x")
f <<= nm.mcal(c='if($s{node1}==$s{node1x},$s{frequency1},$s{frequency2})',a="freq1")
f <<= nm.mcal(c='if($s{node2}==$s{node2x},$s{frequency2},$s{frequency1})',a="freq2")
f <<= nm.mcut(f="i1,i2,frequency,sim,node1x:node1,node2x:node2,freq1:frequency1,freq2:frequency2")
f <<= nm.msetstr(v=total,a="total")
f <<= nm.mcal(c='${frequency}/${frequency1}',a="confidence")
f <<= nm.mcal(c='${frequency}/${total}',a="support")
f <<= nm.mcal(c='${frequency}/(${frequency1}+${frequency2}-${frequency})',a="jaccard")
f <<= nm.mcal(c='(${frequency}*${total})/((${frequency1}*${frequency2}))',a="lift")
f <<= nm.mcal(c='(ln(${frequency})+ln(${total})-ln(${frequency1})-ln(${frequency2}))/(ln(${total})-ln(${frequency}))',a="PMI")
f <<= nm.mcut(f="node1,node2,frequency,frequency1,frequency2,total,support,confidence,lift,jaccard,PMI")
f <<= nm.msortf(f="node1,node2",o=self.oeFile)
f.run()
if self.onFile:
f4 = nm.mcut(f=self.itemFN+":node",i=self.iFile)
f4 <<= nm.mcount(k="node",a="frequency")
if self.node_support :
minstr = "[%s,]"%(minSupp)
f4 <<= nm.mselnum(f="frequency",c=minstr)
f4 <<= nm.msetstr(v=total,a="total")
f4 <<= nm.mcal(c='${frequency}/${total}',a="support")
f4 <<= nm.mcut(f="node,support,frequency,total",o=self.onFile)
f4.run()
procTime=datetime.now()-t
# ログファイル出力
if self.logFile :
kv=[["key","value"]]
for k,v in self.args.items():
kv.append([k,str(v)])
kv.append(["time",str(procTime)])
nm.writecsv(i=kv,o=self.logFile).run()
xxx=[]
n1 = nm.mselstr(i=data1,f="a",v="1")
n3 = n1.msetstr(a="d1",v="x0",o=xxx)
n2 = n1.redirect("u").msetstr(a="d2",v="x1").mproduct(m=n3,f="d1")
print(xxx)
#import nysol.mod as nm
#data1 =[["a","b","c"],["1","2","3"],["4","5","6"]]
n1 = None
n1 <<= nm.mselstr(i=data1,f="a",v="1",o="t1",u="t2")
n1 <<= nm.msetstr(a="add",v="vvv")
print(n1.run())
n2 = None
n2 <<= nm.mcut(i=data1,f="a,b,c")
n2 <<= nm.msum(f="c",o="t4")
n2 <<= nm.mcut(f="b,c",o="t5")
print(n2.run(msg="on"))
xxx1 = []
xxx2 = []
n1 = None
n1 <<= nm.mselstr(i=data1,f="a",v="1",o=xxx1,u=xxx2)
n1 <<= nm.msetstr(a="add",v="vvv")
n1.run()