minimum viable classifier version

neighbours/__init__.py

@@ -1 +1,3 @@
 from .knn_classifier import *
+from . import distance
+from . import kernel

neighbours/distance.py

@@ -0,0 +1,13 @@
+import numpy as np
+
+
+def euclidean(x: np.ndarray, y: np.ndarray):
+    return np.linalg.norm(x - y)
+
+
+def manhattan(x: np.ndarray, y: np.ndarray):
+    return np.sum(np.abs(x - y))
+
+
+def cosine(x: np.ndarray, y: np.ndarray):
+    return 1 - (x.dot(y) / (np.linalg.norm(x) * np.linalg.norm(y)))

neighbours/kernel.py

@@ -0,0 +1,13 @@
+import numpy as np
+
+
+def rectangular(x):
+    return (np.abs(x) <= 1) / 2
+
+
+def gaussian(x):
+    return 0.3989422804014326779 * np.exp(-2 * x * x)
+
+
+def epanechnikov(x):
+    return (np.abs(x) <= 1) * 0.75 * (1 - (x * x))

neighbours/knn_classifier.py

@@ -5,10 +5,11 @@ from .exceptions import *
 
 
 class KNNClassifier:
-    def __init__(self, features, trees_count, rpt_m):
+    def __init__(self, features, classes_count, trees_count, rpt_m):
         self.features = features
         self.forest = RPTForest(features, trees_count, rpt_m)
         self.classes = None
+        self.classes_count = classes_count
 
     def load(self, points, classes):
         if not isinstance(points, np.ndarray):
@@ -29,5 +30,12 @@ class KNNClassifier:
 
         self.forest.load(points)
 
-    def predict(self, point: np.ndarray):
-        pass
+    def predict(self, point: np.ndarray, distance, kernel, h):
+        nearest_point_indexes = self.forest.get_neighbours(point)
+
+        votes = np.zeros(self.classes_count)
+
+        for point_ix in nearest_point_indexes:
+            votes[self.classes[point_ix]] += kernel(distance(point, self.forest.get_point(point_ix)) / h)
+
+        return np.argmax(votes)

neighbours/rp_neighbours.py

@@ -68,7 +68,7 @@ class RPTForest:
     """
 
     def __init__(self, features, trees_count, m):
-        """Creates new RPTForest
+        """Creates new random projection tree forest
 
         :param features: number of features in each sample
         :param trees_count: number of trees in the forest