Iterative kNN with no flash

learners/knn.py

@@ -0,0 +1,23 @@
+from sklearn.neighbors import KNeighborsClassifier
+import numpy as np
+
+
+class kNN(object):
+
+    def __init__(self, classes):
+        self.model = KNeighborsClassifier(n_neighbors=2)
+        self.X = None
+        self.y = None
+
+    def partial_fit(self, x, y):
+        if self.X is None and self.y is None:
+            self.X = np.array([x])
+            self.y = y
+        else:
+            self.X = np.vstack((self.X, x))
+            self.y = np.hstack((self.y, y))
+
+        self.model.fit(self.X, self.y)
+
+    def predict(self, x):
+        return self.model.predict(x)[0]

main.py

@@ -1,21 +1,23 @@
-from sklearn.datasets import load_svmlight_file
-import numpy as np
 from random import shuffle
+import warnings
+
+from sklearn.datasets import load_svmlight_file
+from utils.experiment import test
+
 from ensemblers.adaboost import AdaBooster
 from ensemblers.ogboost import OGBooster
 from ensemblers.ocpboost import OCPBooster
 from ensemblers.expboost import EXPBooster
+
 from learners.sk_naive_bayes import NaiveBayes
 from learners.perceptron import Perceptron
 from learners.random_stump import RandomStump
 from learners.decision_stump import DecisionStump
-from learners.ce_knn import kNN
+from learners.knn import kNN
 from learners.histogram import RNB
 from learners.winnow import Winnow
-from utils.experiment import test
-import warnings
-warnings.filterwarnings("ignore", module="sklearn")
 
+warnings.filterwarnings("ignore", module="sklearn")
 
 def loadData(filename):
     X, y = load_svmlight_file(filename)

utils/test.py

@@ -1,23 +1,33 @@
 import unittest
-from perceptron import Perceptron
+import sys
+sys.path.append("../")
+import numpy as np
+from ..learners.knn import kNN
 
 
-class TestPerceptron(unittest.TestCase):
+class TestLearners(unittest.TestCase):
 
-    def testSimple(self):
+    def setUp(self):
         x1 = [1, 2, 3]
         y1 = 1
         x2 = [1, 3, 5]
-        y2 = -1
+        y2 = 1
         x3 = [-1, -1, -1]
-        y3 = 1
+        y3 = -1
         x4 = [1, 2, 4]
         y4 = 1
-        data = [(x1, y1), (x2, y2), (x3, y3), (x4, y4)]
-        p = Perceptron()
-        for (x, y) in data:
-            p.update(x, y)
-        self.assertEqual(p.w, [1.0, 1.0, 2.0])
+        x5 = [-0.5, -0.5, 0]
+        y5 = -1
+        data = [(x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5)]
+        self.data = [(np.array(x), np.array([y])) for (x, y) in data]
+        self.labels = np.array([-1, 1])
+
+    def testKNN(self):
+        model = kNN(self.labels)
+        for (x, y) in self.data:
+            model.partial_fit(x, y)
+        print model.predict(np.array([0, 0, 0]))
+
 
 if __name__ == '__main__':
     unittest.main()