def __work():
print "Reading", filename
count, dicts = mushroom_dicts(filename)
seed = shuffle.generate_seed(count)
o = mushroom(filename, count, dicts, seed)
pickle.dump(o, open(filename + ".bin", "w"))
print "Written", filename + ".bin"
def optdigits(filename):
fh = open(filename, "r")
data = []
labels = []
fl = fh.readlines()
for line in fl:
vals = line.split('\n')[0].rsplit(',')
if len(vals) != 65:
break
data.append([int(vals[i]) for i in range(64)])
labels.append(vals[64])
shuffle.shuffle(data, labels, shuffle.generate_seed(len(data)))
return numpy.array(data), labels
def glass(filename):
fh = open(filename, "r")
data = []
labels = []
fl = fh.readlines()
for line in fl:
vals = line.split('\n')[0].rsplit(',')
if len(vals) != 11:
break
data.append([float(vals[i + 1]) for i in range(9)])
labels.append(tr(vals[10]))
shuffle.shuffle(data, labels, shuffle.generate_seed(len(data)))
return numpy.array(data), labels