Implementation of K-NN approach take suitable example using python

filename: b13.py

import csv
from math import sqrt
import random
from collections import Counter
from operator import itemgetter

dataset = []
training = []
test = []

def loaddata():
 with open('iris.data','rb') as csvfile:
  lines = csv.reader(csvfile)
  dataset = list(lines)
  random.shuffle(dataset)
  for x in range(15):
   training.append(dataset[x])
  for x in range(15,20):
   test.append(dataset[x])

def calculate():
 for x in test:
  distances = []
  abc = []
  for y in training:
   d = pow(int(x[0])-int(y[0]),2)
   distances.append([sqrt(d),y])
  sorteddistance = sorted(distances,key=itemgetter(0))
  topk = sorteddistance[:3]
  for (a,b) in topk:
   abc.append(b)
  classes = Counter(t for(p,t) in abc)
  print "Class for"+str(x)+':'+ max(classes, key=lambda cls:classes[cls])

def main():
 loaddata()
 calculate()

main()

filename: iris.data

10,A
20,A
30,A
40,A
50,A
60,B
70,B
80,B
90,B
100,B
110,C
120,C
130,C
140,C
150,c
22,A
62,B
92,B
11,A
124,C

OUTPUT

ccpvg@ccpvg-HP-Compaq-4000-Pro-SFF-PC:~$python B13.py
Class for['40', 'A']:A
Class for['60', 'B']:B
Class for['92', 'B']:B
Class for['124', 'C']:C
Class for['90', 'B']:B