digit_recognition.py

import numpy as np
import matplotlib.pyplot as pt
import pandas as pd
from sklearn.tree import DecisionTreeClassifier

data = pd.read_csv("train.csv").as_matrix()
clf = DecisionTreeClassifier()

#training data
xtrain = data[0:21000, 1:]
train_label = data[0:21000, 0]

clf.fit(xtrain,train_label)

#testing data
xtest = data[21000:, 1:]
actual_label = data[21000:, 0]

#======================================================================== 
# To plot the particular value and print the same.
# d = xtest[9]
# d.shape= (28,28)  #plotting the data so that we can see if it matches.
# pt.imshow(d,cmap='gray')
# pt.show()

# print (clf.predict([xtest[9]]))
#========================================================================
p = clf.predict(xtest)
count = 0
for i in range(0,21000):
    count +=1 if p[i] == actual_label[i] else 0
print ("Accuracy = {}".format((count/21000)*100))