diff --git a/src/ensemble/__pycache__/boosting.cpython-310.pyc b/src/ensemble/__pycache__/boosting.cpython-310.pyc
index 526d2d1..1a2bf45 100644
Binary files a/src/ensemble/__pycache__/boosting.cpython-310.pyc and b/src/ensemble/__pycache__/boosting.cpython-310.pyc differ
diff --git a/src/ensemble/boosting.py b/src/ensemble/boosting.py
index c7abdc8..d060b68 100644
--- a/src/ensemble/boosting.py
+++ b/src/ensemble/boosting.py
@@ -1,4 +1,5 @@
 from abc import abstractmethod
+
 import numpy as np
 
 from src.base import Model
@@ -8,7 +9,7 @@
 class _GradientBoosting(Model):
     def __init__(
             self,
-            learning_rate: float = 1e-3,
+            learning_rate: float = 1e-2,
             n_steps: int = 100,
             max_depth: int = 3,
             min_samples_split: int = 2,
@@ -77,19 +78,21 @@ def _calculate_loss_gradient(self, y: np.ndarray, predictions: np.ndarray) -> np
         """
         pass
 
+    @abstractmethod
+    def _calculate_predictions(self, x: np.ndarray) -> np.ndarray:
+        """
+        Calculate predictions for input data.
+        :param x: Input data.
+        :return: Predictions.
+        """
+
     def predict(self, x: np.ndarray) -> np.ndarray:
         """
         Predict target feature using pretrained boosting trees.
         :param x: Test data.
         :return: Test predictions.
         """
-        n_samples, _ = x.shape
-
-        predictions = np.ones(shape=(n_samples,)) * self.constant_prediction
-
-        for tree in self.trees:
-            predictions = predictions + self.learning_rate * tree.predict(x)
-
+        predictions = self._calculate_predictions(x)
         return predictions
 
 
@@ -108,11 +111,75 @@ def _calculate_initial_prediction(self, y: np.ndarray) -> np.ndarray:
 
     def _calculate_loss_gradient(self, y: np.ndarray, predictions: np.ndarray) -> np.ndarray:
         """
-        Find mean value for the targets.
+        Calculate gradient of mean-squared error loss.
+        :param predictions: Target predictions.
+        :param y: Targets.
+        :return: Gradient of loss function with respect to predictions.
+        """
+        return y - predictions
+
+    def _calculate_predictions(self, x: np.ndarray) -> np.ndarray:
+        n_samples, _ = x.shape
+
+        predictions = np.ones(n_samples) * self.constant_prediction
+        for tree in self.trees:
+            predictions = predictions + self.learning_rate * tree.predict(x)
+
+        return predictions
+
+
+class GradientBoostingClassifier(_GradientBoosting):
+    """
+    Gradient Boosting for the classification.
+    Uses cross-entropy as loss.
+    """
+
+    def _calculate_initial_prediction(self, y: np.ndarray) -> np.ndarray:
+        """
+        Find natural logarithm of odds.
         :param y: Targets.
         :return: Initial predictions.
         """
-        return predictions - y
+        return np.zeros_like(y, dtype=np.float64)
+
+    def _calculate_loss_gradient(self, y: np.ndarray, predictions: np.ndarray) -> np.ndarray:
+        """
+        Calculate cross-entropy gradient.
+        :param y: Targets.
+        :return: Gradient of loss function with respect to predictions.
+        """
+        return y - GradientBoostingClassifier.sigmoid(predictions)
 
+    @staticmethod
+    def sigmoid(x: np.ndarray) -> np.ndarray:
+        """
+        Makes input values to be in (0, 1) range.
+        :param x: Input array.
+        :return: Output array of the same shape as an input array.
+        """
+        return 1 / (1 + np.exp(-x))
+
+    def _calculate_predictions(self, x: np.ndarray) -> np.ndarray:
+        """
+        Calculate targets using prediction probability.
+        :param x: Input array.
+        :return: Predictions.
+        """
+        predictions_proba = self.predict_proba(x)
+        predictions = np.where(predictions_proba >= 0.5, 1, 0)
+        return predictions
+
+    def predict_proba(self, x):
+        """
+        Predict label using sigmoid function.
+        :param x: Input array.
+        :return: Predictions.
+        """
+        n_samples, _ = x.shape
+
+        predictions = np.ones(n_samples) * self.constant_prediction
+        for tree in self.trees:
+            predictions = predictions + self.learning_rate * tree.predict(x)
 
+        return GradientBoostingClassifier.sigmoid(predictions)
 
diff --git a/tests/base/__pycache__/config.cpython-310.pyc b/tests/base/__pycache__/config.cpython-310.pyc
index ea5d843..cd81cb2 100644
Binary files a/tests/base/__pycache__/config.cpython-310.pyc and b/tests/base/__pycache__/config.cpython-310.pyc differ
diff --git a/tests/base/config.py b/tests/base/config.py
index 9648eba..01ec3b8 100644
--- a/tests/base/config.py
+++ b/tests/base/config.py
@@ -55,6 +55,7 @@ def check_fit_predict(model, x: np.ndarray, y: np.ndarray):
     # Fit the model on the mock dataset
     model.fit(x, y)
     preds = model.predict(x)
+    print(preds)
 
     assert isinstance(preds, np.ndarray)
     assert preds.shape == y.shape
diff --git a/tests/ensemble/__pycache__/test_boosting.cpython-310-pytest-7.4.4.pyc b/tests/ensemble/__pycache__/test_boosting.cpython-310-pytest-7.4.4.pyc
index c51f559..56cf065 100644
Binary files a/tests/ensemble/__pycache__/test_boosting.cpython-310-pytest-7.4.4.pyc and b/tests/ensemble/__pycache__/test_boosting.cpython-310-pytest-7.4.4.pyc differ
diff --git a/tests/ensemble/test_boosting.py b/tests/ensemble/test_boosting.py
index 52b0eb3..3331fe8 100644
--- a/tests/ensemble/test_boosting.py
+++ b/tests/ensemble/test_boosting.py
@@ -1,8 +1,14 @@
-from src.ensemble import GradientBoostingRegressor
-from tests.base.config import dataset_regression, check_fit_predict  # noqa: F401
+from src.ensemble import GradientBoostingRegressor, GradientBoostingClassifier
+from tests.base.config import dataset_regression, dataset_classification, check_fit_predict  # noqa: F401
 
 
 def test_gradient_boosting_regression(dataset_regression):
     x, y = dataset_regression
-    model = GradientBoostingRegressor()
+    model = GradientBoostingRegressor(learning_rate=5e-2)
+    check_fit_predict(model, x, y)
+
+
+def test_gradient_boosting_classification(dataset_classification):
+    x, y = dataset_classification
+    model = GradientBoostingClassifier()
     check_fit_predict(model, x, y)