# model=SVC(C=c,gamma=g)
# print('Fitting model with params c: %f, g: %f ...'%(c,g))
# model=model.fit(trset,trlabels)
# try:
# scorei=model.score(cvset,cvlabels)
# except:
# scorei=0 # something went wrong...
# scores[counter]=np.mean(scorei)
# paramholder[counter,0]=c
# paramholder[counter,1]=g
# if scorei>bestscore:
# bestscore=scorei
# bestmodel=model
# counter+=1
# print('Score = %f with c: %f, g: %f' %(scorei,c,g))
#bestc=paramholder[scores.argmax(),0]
#bestg=paramholder[scores.argmax(),1]
#print('Best score of %f with c: %f, g: %f' %(bestscore,bestc,bestg))
print('Fitting model with params c: %f, g: %f ...'%(c,g))
model=SVC(C=c,gamma=g)
model=model.fit(trset,trlabels)
pickle.dump(model,open('data/svm_c5_g05.p','wb'))
# predict test set
testx=imt.loadimgfromcsv(testfile,colstart=0)/255.0
ntest=testx.shape[0]
preds = model.predict(testx)
# output to .csv file labels
imt.writeoutpred(map(int,preds),np.arange(ntest)+1,outfile,header=["ImageId","Label"])
nx = x.shape[0]
for zz in range(nx):
sdAE.GD(x[zz].reshape(x.shape[1], 1), labels[lcount], alpha=0.1)
lcount += 1
iter += 1
# check model with CV set
cvscore = sdAE.score(cvset, cvlabels)
if cvscore >= bestscore:
# best so far, but wait a bit
if cvscore >= bestscore * improvethresh:
# increase patience
patience = np.maximum(patience, iter * patienceinc)
bestscore = cvscore
if iter >= patience:
# done!
loopflag = True
print('Exiting optimization. CV score= %3f' % cvscore)
break # out of block loop
ep += 1
# pickle the best model
pickle.dump(sdAE, open('data/finetuned.p', 'wb'))
# predict labels from test dataset
# process in all at once
testx = imt.loadimgfromcsv(testfile, colstart=0) / 255.0
ntest = testx.shape[0]
preds = sdAE.predict(testx.T)
# output to .csv file labels
imt.writeoutpred(preds, range(ntest) + 1, outfile, header=["ImageId", "Label"])
# model=model.fit(trset,trlabels)
# try:
# scorei=model.score(cvset,cvlabels)
# except:
# scorei=0 # something went wrong...
# scores[counter]=np.mean(scorei)
# paramholder[counter,0]=c
# paramholder[counter,1]=g
# if scorei>bestscore:
# bestscore=scorei
# bestmodel=model
# counter+=1
# print('Score = %f with c: %f, g: %f' %(scorei,c,g))
#bestc=paramholder[scores.argmax(),0]
#bestg=paramholder[scores.argmax(),1]
#print('Best score of %f with c: %f, g: %f' %(bestscore,bestc,bestg))
print('Fitting model with params c: %f, g: %f ...' % (c, g))
model = SVC(C=c, gamma=g)
model = model.fit(trset, trlabels)
pickle.dump(model, open('data/svm_c5_g05.p', 'wb'))
# predict test set
testx = imt.loadimgfromcsv(testfile, colstart=0) / 255.0
ntest = testx.shape[0]
preds = model.predict(testx)
# output to .csv file labels
imt.writeoutpred(map(int, preds),
np.arange(ntest) + 1,
outfile,
header=["ImageId", "Label"])
nx=x.shape[0]
for zz in range(nx):
sdAE.GD(x[zz].reshape(x.shape[1],1),labels[lcount],alpha=0.1)
lcount+=1
iter+=1
# check model with CV set
cvscore=sdAE.score(cvset,cvlabels)
if cvscore >= bestscore:
# best so far, but wait a bit
if cvscore >= bestscore*improvethresh:
# increase patience
patience = np.maximum(patience,iter*patienceinc)
bestscore = cvscore
if iter >= patience:
# done!
loopflag=True
print('Exiting optimization. CV score= %3f'%cvscore)
break # out of block loop
ep+=1
# pickle the best model
pickle.dump(sdAE,open('data/finetuned.p','wb'))
# predict labels from test dataset
# process in all at once
testx=imt.loadimgfromcsv(testfile,colstart=0)/255.0
ntest=testx.shape[0]
preds = sdAE.predict(testx.T)
# output to .csv file labels
imt.writeoutpred(preds,range(ntest)+1,outfile,header=["ImageId","Label"])
