class WekaModel:
def __init__(self, weka_model="weka.classifiers.trees.J48",addops=[]):
self.weka_model=weka_model
self.addops=addops
self.tmpdir=TempDirectory()
@staticmethod
def load(file_, *args, **kwargs):
assert(False)
def train(self,data,label,online=False,cross_validation_result=True):
assert(online==False)
#print "DATA=",data
#print "LABEL=",label
infile=os.path.join(self.tmpdir.get(),"train.arff")
mdlfile=os.path.join(self.tmpdir.get(),"out.mdl")
addops=["-r", "-v"]
if cross_validation_result==False:
addops+=["-no-cv" ]
xdata=numpy.hstack([data,label])
numpy2arff(infile,xdata)
cmdline=' '.join(["CLASSPATH='%s'"%(WEKACLASSPATH,),JAVA_CMD,self.weka_model, "-t",infile , "-d",mdlfile]+self.addops+addops)
print cmdline
os.system(cmdline)
def test(self,test_data,ground_truth=None):
assert(online==False)
infile=os.path.join(self.tmpdir.get(),"test.arff")
mdlfile=os.path.join(self.tmpdir.get(),"out.mdl")
addops=[]
numpy2arff(infile,test_data)
os.system(' '.join(["CLASSPATH='%s'"%(WEKACLASSPATH,),JAVA_CMD,self.weka_model, "-l",mdlfile, "-T",infile ]))
def memory_cost(self, *args, **kwargs):
assert(False)
def cpu_cost(self, *args, **kwargs):
assert(False)
def random_improve(self,value,amount=0.5, prec=1):
## TRY TO RANDOMLY CHANGE WITH A NEAREST NEIGHBOR
#print "random_improve not yet implemented doing sample instead !!! hist"
#return self.sample(value.shape[0])
assert(False)
def __init__(self, weka_model="weka.classifiers.trees.J48",addops=[]):
self.weka_model=weka_model
self.addops=addops
self.tmpdir=TempDirectory()
