knn

"""1. KNN on Iris Datset.ipynb

from sklearn.datasets import load_iris

#Create bunch object containing iris dataset and its attributes.
iris = load_iris()


X = iris.data
y = iris.target

from sklearn.cross_validation import train_test_split
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.2,random_state=4)

from sklearn.neighbors import KNeighborsClassifier

from sklearn import metrics
#Try running from k=1 through 25 and record testing accuracy
k_range = range(1,26)
scores = {}
scores_list = []
for k in k_range:
        knn = KNeighborsClassifier(n_neighbors=k)
        knn.fit(X_train,y_train)
        y_pred=knn.predict(X_test)
        scores[k] = metrics.accuracy_score(y_test,y_pred)
        scores_list.append(metrics.accuracy_score(y_test,y_pred))

#Testing accuracy for each value of K
scores

# Commented out IPython magic to ensure Python compatibility.

# %matplotlib inline
import matplotlib.pyplot as plt

#plot the relationship between K and the testing accuracy
plt.plot(k_range,scores_list)
plt.xlabel('Value of K for KNN')
plt.ylabel('Testing Accuracy')

"""### K values with 3 to 19 has the same accuracy which is 96.66, so we can use any one value from that, i choose K as 5 and train the model with full training data"""

knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(X,y)

#0 = setosa, 1=versicolor, 2=virginica
classes = {0:'setosa',1:'versicolor',2:'virginica'}

#Making prediction on some unseen data
#predict for the below two random observations
x_new = [[3,4,5,2],
         [5,4,2,2]]
y_predict = knn.predict(x_new)

print(classes[y_predict[0]])
print(classes[y_predict[1]])