Create knn.py

knn.py

@@ -0,0 +1,69 @@
+__author__ = 'sravan'
+from numpy import *
+import operator
+
+def calcDistance(obj1,obj2):
+    x1 = int(obj1[1])
+    y1 = int(obj1[2])
+    x2 = int(obj2[1])
+    y2 = int(obj2[2])
+    #print x1,x2,y1,y2
+    dist = sqrt( (x2 - x1)**2 + (y2 - y1)**2 )
+    #print dist
+    return dist
+
+def getLabel(dataset,index):
+    #print dataset,index
+    #iterating dataset till index and returning its label
+    obj = dataset[index]
+    label = obj[0]
+    return label
+
+def createDataSet():
+    group = [['A',1,22],['A',1,18],['B',22,0],['A',1,17],['A',1,18],['B',25,0],['B',27,0.1],['C',12,5],['C',12,4]]
+    return group
+
+def getMaxOccurLabel(label_list):
+    label_dict={}
+    for label in label_list:
+        if label_dict.has_key(label)==False:
+            label_dict[label]=1
+        else:
+            label_dict[label]=label_dict[label]+1
+    sorted_x = sorted(label_dict.iteritems(), key=operator.itemgetter(1), reverse=True)
+    max_occur_label = sorted_x[0][0]
+    return max_occur_label
+
+def knn(dataset,obj,k):
+    if k>len(dataset):
+        print 'Dataset size : '+str(len(dataset))
+        print 'K should be less than dataset size, given K size as : '+str(k)
+        return False
+    distances={}
+    sorted_x={}
+    i=0
+    for item in dataset:
+        #print item
+        dist = calcDistance(item,obj)
+        distances[i] = int(dist)
+        i+=1
+    print distances
+    #will have index and distance from object(index,distance)
+    sorted_x = sorted(distances.iteritems(), key=operator.itemgetter(1))
+    #print sorted_x
+    cnt=0
+    labels_list=[]
+    #index,distance
+    while cnt<k:
+        tupple = sorted_x[cnt]
+        index = tupple[0]
+        label = getLabel(dataset,index)
+        labels_list.append(label)
+        #print label
+        cnt+=1
+    print getMaxOccurLabel(labels_list)
+    return
+
+dataset = createDataSet()
+obj=['unknown',1,5]
+knn(dataset,obj,3)