@@ -128,6 +128,13 @@ Support for Python 3.4 and below has been officially dropped.
128128 :class: `ensemble.AdaBoostClassifier ` and :class: `ensemble.AdaBoostRegressor `
129129 :issue: `13174 ` by :user: `Christos Aridas <chkoar> `.
130130
131+ - |Fix | Fixed a bug in :class: `ensemble.GradientBoostingClassifier ` and
132+ :class: `ensemble.GradientBoostingRegressor `, which didn't support
133+ scikit-learn estimators as the initial estimator. Also added support of
134+ initial estimator which does not support sample weights. :issue: `12436 ` by
135+ :user: `Jérémie du Boisberranger <jeremiedbb> ` and :issue: `12983 ` by
136+ :user: `Nicolas Hug<NicolasHug> `.
137+
131138- |Fix | Fixed the output of the average path length computed in
132139 :class: `ensemble.IsolationForest ` when the input is either 0, 1 or 2.
133140 :issue: `13251 ` by :user: `Albert Thomas <albertcthomas> `
@@ -137,6 +144,11 @@ Support for Python 3.4 and below has been officially dropped.
137144 the gradients would be incorrectly computed in multiclass classification
138145 problems. :issue: `12715 ` by :user: `Nicolas Hug<NicolasHug> `.
139146
147+ - |Fix | Fixed a bug in :class: `ensemble.GradientBoostingClassifier ` where
148+ the default initial prediction of a multiclass classifier would predict the
149+ classes priors instead of the log of the priors. :issue: `12983 ` by
150+ :user: `Nicolas Hug<NicolasHug> `.
151+
140152- |Fix | Fixed a bug in :mod: `ensemble ` where the ``predict `` method would
141153 error for multiclass multioutput forests models if any targets were strings.
142154 :issue: `12834 ` by :user: `Elizabeth Sander <elsander> `.
0 commit comments