6faa0f28ef2fd974ec5e527582291434d1182952,bugbug/models/tracking.py,TrackingModel,__init__,#TrackingModel#Any#,18

Before Change

            bug_features.cleanup_synonyms,
        ]

        self.data_vectorizer = DictVectorizer()
        self.title_vectorizer = self.text_vectorizer(stop_words="english")
        self.comments_vectorizer = self.text_vectorizer(stop_words="english")

        self.extraction_pipeline = Pipeline([
            ("bug_extractor", bug_features.BugExtractor(feature_extractors, cleanup_functions, rollback=True, rollback_when=self.rollback)),
            ("union", FeatureUnion(
                transformer_list=[
                    ("data", Pipeline([
                        ("selector", DictSelector(key="data")),
                        ("vect", self.data_vectorizer),
                    ])),

                    ("title", Pipeline([
                        ("selector", DictSelector(key="title")),
                        ("tfidf", self.title_vectorizer),
                    ])),

                    ("comments", Pipeline([
                        ("selector", DictSelector(key="comments")),
                        ("tfidf", self.comments_vectorizer),
                    ])),
                ],
            )),
        ])

        self.clf = xgboost.XGBClassifier(n_jobs=16)

After Change

            bug_features.cleanup_synonyms,
        ]

        self.extraction_pipeline = Pipeline([
            ("bug_extractor", bug_features.BugExtractor(feature_extractors, cleanup_functions, rollback=True, rollback_when=self.rollback)),
            ("union", ColumnTransformer([
                ("data", DictVectorizer(), "data"),

                ("title", self.text_vectorizer(stop_words="english"), "title"),

                ("comments", self.text_vectorizer(stop_words="english"), "comments"),
            ])),
        ])

        self.clf = xgboost.XGBClassifier(n_jobs=16)

Instances