test commit

homeworks/assignment01_knn/k_nearest_neighbor.py

@@ -75,7 +75,7 @@ def compute_distances_two_loops(self, X):
                 # not use a loop over dimension, nor use np.linalg.norm().          #
                 #####################################################################
                 # *****START OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
-
+                dists[i, j] = np.sqrt(np.sum((X[i] - self.X_train[j]) ** 2))
                 # *****END OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
         return dists
 
@@ -97,7 +97,7 @@ def compute_distances_one_loop(self, X):
             # Do not use np.linalg.norm().                                        #
             #######################################################################
             # *****START OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
-
+            dists[i, :] = np.sqrt(np.sum((X[i] - self.X_train) ** 2, axis=1))
             # *****END OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
         return dists
 
@@ -126,6 +126,12 @@ def compute_distances_no_loops(self, X):
         #########################################################################
         # *****START OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
 
+        test_squared = np.sum(X ** 2, axis=1, keepdims=True)
+        train_squared = np.sum(self.X_train ** 2, axis=1)
+        inner_product = np.dot(X, self.X_train.T)
+
+        # Вычислите квадрат расстояния Евклида между всеми парами точек
+        dists = np.sqrt(test_squared - 2 * inner_product + train_squared)
         # *****END OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
         return dists
 
@@ -156,6 +162,10 @@ def predict_labels(self, dists, k=1):
             #########################################################################
             # *****START OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
 
+            closest_k_indices = np.argsort(dists[i])[:k]
+            closest_y = self.y_train[closest_k_indices]
+
+
             # *****END OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
             #########################################################################
             # TODO:                                                                 #
@@ -166,6 +176,8 @@ def predict_labels(self, dists, k=1):
             #########################################################################
             # *****START OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
 
+            counts = np.bincount(closest_y)
+            y_pred[i] = np.argmax(counts)
 
             # *****END OF YOUR CODE (DO NOT DELETE/MODIFY THIS LINE)*****
 

homeworks/assignment02_laplace/distribution.py

@@ -11,17 +11,21 @@ def mean_abs_deviation_from_median(x: np.ndarray):
         ####
         # Do not change the class outside of this block
         # Your code here
+        median = np.median(x)  # Находим медиану для каждой фичи
+        deviation = np.abs(x - median)  # Находим абсолютное отклонение от медианы
+        mean_deviation = np.mean(deviation)  # Находим среднее абсолютное отклонение
+        return mean_deviation
         ####
 
-    def __init__(self, features):
+    def __init__(self, features, loc=0.0, scale=1.0):
         '''
         Args:
             feature: A numpy array of shape (n_objects, n_features). Every column represents all available values for the selected feature.
+            loc: The location (mean) parameter of the Laplace distribution.
+            scale: The scale (standard deviation) parameter of the Laplace distribution.
         '''
-        ####
-        # Do not change the class outside of this block
-        self.loc = # YOUR CODE HERE
-        self.scale = # YOUR CODE HERE
+        self.loc = loc
+        self.scale = scale
         ####
 
 
@@ -33,7 +37,10 @@ def logpdf(self, values):
         '''
         ####
         # Do not change the class outside of this block
-        return 
+        exponent = -np.abs(values - self.loc) / self.scale
+        log_density = -np.log(2 * self.scale) + exponent
+        return log_density
+
         ####