Refactored 'DiscreteDomain.fit(X, y)' method

sklearn2pmml/decoration/__init__.py

@@ -185,13 +185,12 @@ def _isin_mask(x, values):
 				mask[where] = _isin_mask(X[where], data_values)
 				return mask
 			else:
-				if isinstance(data_values, list):
-					if X.shape[1] != len(data_values):
-						raise ValueError()
+				if hasattr(data_values, "__len__"):
+					_check_cols(X, data_values)
 				mask = numpy.full(X.shape, fill_value = False)
 				for col in range(X.shape[1]):
 					col_where = where[:, col]
-					mask[col_where, col] = _isin_mask(X[col_where, col], data_values[col] if isinstance(data_values, list) else data_values)
+					mask[col_where, col] = _isin_mask(X[col_where, col], data_values[col] if hasattr(data_values, "__len__") else data_values)
 				return mask
 		return super(DiscreteDomain, self)._valid_value_mask(X, where)
 
@@ -202,50 +201,47 @@ def fit(self, X, y = None):
 		if self._empty_fit():
 			return self
 		X = to_numpy(X)
-		missing_mask = self._missing_value_mask(X)
-		nonmissing_mask = ~missing_mask
-		if is_1d(X):
-			if self.with_statistics:
-				values, counts = numpy.unique(X[nonmissing_mask], return_counts = True)
-			else:
-				values = numpy.unique(X[nonmissing_mask])
-			if self.with_data:
+		if self.with_data:
+			missing_mask = self._missing_value_mask(X)
+			nonmissing_mask = ~missing_mask
+			if is_1d(X):
 				if _is_pandas_categorical(self.dtype_):
 					data_values = self.dtype_.categories
 				else:
-					data_values = values
+					data_values = numpy.unique(X[nonmissing_mask])
 				if (self.missing_value_replacement is not None) and numpy.any(missing_mask) > 0:
 					if _is_pandas_categorical(self.dtype_):
 						raise ValueError()
 					data_values = numpy.unique(numpy.append(data_values, self.missing_value_replacement))
 				self.data_values_ = data_values
-			if self.with_statistics:
-				self.counts_ = _count(missing_mask, nonmissing_mask, None)
-				self.discr_stats_ = (values, counts)
-		else:
-			if self.with_data:
+			else:
+				if _is_pandas_categorical(self.dtype_):
+					raise ValueError()
 				self.data_values_ = []
-			if self.with_statistics:
-				self.counts_ = []
-				self.discr_stats_ = []
-			for col in range(X.shape[1]):
-				col_X = X[:, col]
-				col_missing_mask = missing_mask[:, col]
-				col_nonmissing_mask = nonmissing_mask[:, col]
-				if self.with_statistics:
-					values, counts = numpy.unique(col_X[col_nonmissing_mask], return_counts = True)
-				else:
-					values = numpy.unique(col_X[col_nonmissing_mask])
-				if self.with_data:
-					if _is_pandas_categorical(self.dtype_):
-						raise ValueError()
-					else:
-						data_values = values
+				for col in range(X.shape[1]):
+					col_X = X[:, col]
+					col_missing_mask = missing_mask[:, col]
+					col_nonmissing_mask = nonmissing_mask[:, col]
+					data_values = numpy.unique(col_X[col_nonmissing_mask])
 					if (self.missing_value_replacement is not None) and numpy.any(col_missing_mask) > 0:
 						data_values = numpy.unique(numpy.append(data_values, self.missing_value_replacement))
 					self.data_values_.append(data_values)
-				if self.with_statistics:
-					self.counts_.append(_count(col_missing_mask, ~col_missing_mask, None))
+		if self.with_statistics:
+			missing_mask, valid_mask, invalid_mask = self._compute_masks(X)
+			if is_1d(X):
+				values, counts = numpy.unique(X[valid_mask], return_counts = True)
+				self.counts_ = _count(missing_mask, valid_mask, invalid_mask)
+				self.discr_stats_ = (values, counts)
+			else:
+				self.counts_ = []
+				self.discr_stats_ = []
+				for col in range(X.shape[1]):
+					col_X = X[:, col]
+					col_missing_mask = missing_mask[:, col]
+					col_valid_mask = valid_mask[:, col]
+					col_invalid_mask = invalid_mask[:, col]
+					values, counts = numpy.unique(col_X[col_valid_mask], return_counts = True)
+					self.counts_.append(_count(col_missing_mask, col_valid_mask, col_invalid_mask))
 					self.discr_stats_.append((values, counts))
 		return self