Fix ZeroDivisionError bug (#10)

When the model was trained using integer numbers as labels, and when one
of the labels was the 0 value, the class with this labels and all its
training documents were ignored. Therefore, if the problem being solved
was a binary class classification, the 0 class was never learned and as
such the model was trained using only a single class which lead to the
ZeroDivisionError.

The fix was really straightforward, the issue was cause by the following
condition in the SS3.learn() method:

    if not doc or not cat:
            return

Replacing the ``not cat`` with ``cat is None`` solved the issue.

Resolves: #10

pyss3/__init__.py

@@ -214,7 +214,10 @@ def __sn__(self, ngram, icat):
 
  s = sum([min(v, 1) for v in m_values])
 
- return pow((c - (s + 1)) / ((c - 1) * (s + 1)), self.__p__)
+ try:
+ return pow((c - (s + 1)) / ((c - 1) * (s + 1)), self.__p__)
+ except ZeroDivisionError: # if c <= 1
+ return 1.
 
  def __sg_vanilla__(self, ngram, icat, cache=True):
  """The original significance (sg) function definition."""
@@ -1936,7 +1939,7 @@ def learn(self, doc, cat, n_grams=1, prep=True, update=True):
  """
  self.__cv_cache__ = None
 
- if not doc or not cat:
+ if not doc or cat is None:
  return
 
  try:

tests/test_pyss3.py

@@ -295,6 +295,19 @@ def test_pyss3_ss3(mockers):
  with pytest.raises(ValueError):
  clf = SS3("hyperparameter")
 
+ # Using integer labels
+ test_x = ["this is the first document"] * 5 + ["this is the second document"] * 5
+ test_y = [0] * 5 + [1] * 5
+ clf = SS3()
+ clf.train(test_x, test_y)
+ assert clf.classify_label("this is the first document") == 0
+ assert clf.classify_label("this is the second document") == 1
+
+ # traning only with one category
+ clf = SS3()
+ clf.train(["this is the first document"], ["first"])
+
+ # training different cases
  clf = SS3(
  s=.45, l=.5, p=1, a=0,
  cv_m=STR_NORM_GV_XAI, sn_m=STR_XAI