def divide_corpus_kfold(corpus, k):
n = sys.maxsize
for label in corpus.cuisines.keys():
c = corpus.cuisines[label]
n = min(n, len(c.recipes))
ret = []
prev_idx = -1
window_sz = math.floor(n / k)
for i in range(0, k):
idx1 = prev_idx + 1
idx2 = window_sz + idx1 - 1
prev_idx = idx2
#print('idx1=%d idx2=%d'%(idx1,idx2));
ctrain = []
ctest = []
for label in corpus.cuisines.keys():
c = corpus.cuisines[label]
rtest = c.recipes[idx1:idx2 + 1]
rtrain = c.recipes[0:idx1] + c.recipes[idx2 + 1:]
ctrain.append(cuisine(label=label, recipes=rtrain))
ctest.append(cuisine(label=label, recipes=rtest))
cctrain = cuisine_corpus(cuisines=ctrain)
cctest = cuisine_corpus(cuisines=ctest)
ret.append([cctrain, cctest])
return ret
def divide_corpus(corpus,n): cuisines = []; for label in corpus.cuisines.keys(): c = corpus.cuisines[label]; recipes = c.select_recipes(n); cuisines.append(cuisine(label=label,recipes=recipes)); dcorpus = cuisine_corpus(cuisines=cuisines); return dcorpus;
def divide_corpus(corpus, n):
cuisines = []
for label in corpus.cuisines.keys():
c = corpus.cuisines[label]
recipes = c.select_recipes(n)
cuisines.append(cuisine(label=label, recipes=recipes))
dcorpus = cuisine_corpus(cuisines=cuisines)
return dcorpus
def divide_corpus_kfold(corpus,k):
n = sys.maxsize;
for label in corpus.cuisines.keys():
c = corpus.cuisines[label];
n = min(n,len(c.recipes));
ret = [];
prev_idx = -1;
window_sz = math.floor(n/k);
for i in range(0,k):
idx1 = prev_idx+1; idx2 = window_sz + idx1-1; prev_idx = idx2;
#print('idx1=%d idx2=%d'%(idx1,idx2));
ctrain = []; ctest = [];
for label in corpus.cuisines.keys():
c = corpus.cuisines[label];
rtest = c.recipes[idx1:idx2+1];
rtrain = c.recipes[0:idx1] + c.recipes[idx2+1:];
ctrain.append(cuisine(label=label,recipes=rtrain));
ctest.append(cuisine(label=label,recipes=rtest));
cctrain = cuisine_corpus(cuisines=ctrain);
cctest = cuisine_corpus(cuisines=ctest);
ret.append([cctrain,cctest]);
return ret;
def get_cuisine(fname):
base = os.path.basename(fname);
stext = os.path.splitext(base);
label = stext[0];
ext = stext[1];
if ext!='.csv':
return None;
print(fname);
with open(fname) as f:
#lines = f.readlines();
lines = csv.reader(f);
c = cuisine(label=label);
i = 0;
for l in lines:
try:
i = i+1;
c.add_recipe(recipe(l));
except Exception as E:
print('lines:%s'%(i));
print(E);
return c;
def get_cuisine(fname):
base = os.path.basename(fname)
stext = os.path.splitext(base)
label = stext[0]
ext = stext[1]
if ext != '.csv':
return None
print(fname)
with open(fname) as f:
#lines = f.readlines();
lines = csv.reader(f)
c = cuisine(label=label)
i = 0
for l in lines:
try:
i = i + 1
c.add_recipe(recipe(l))
except Exception as E:
print('lines:%s' % (i))
print(E)
return c