knn_sklearn_test.py

from __future__ import division
from sklearn.neighbors import KNeighborsClassifier as KNN_sklearn
import timeit
import numpy as np
from orient import KNN
from sklearn.decomposition import PCA

def read_file(fname):
    X = np.loadtxt(fname, usecols=range(2, 194), dtype=int)
    y = np.loadtxt(fname, usecols=1, dtype=int)
    return X/255, y


pca = PCA(n_components=0.85, svd_solver="full")

X, y = read_file("train-data.txt")
X = pca.fit_transform(X)

Xt, yt = read_file("test-data.txt")
Xt = pca.transform(Xt)

tic = timeit.default_timer()

knn = KNN_sklearn()
knn.fit(X, y)
score = knn.score(Xt, yt)
print (round(score, 2)*100, "%")

toc = timeit.default_timer()
print ("Time", int(toc - tic))

print ("===============")

scores = []
for k in range(180, 220, 10):
    tic = timeit.default_timer()

    knn = KNN(k)
    knn.train(X, y)
    score = knn.test(Xt, yt)

    toc = timeit.default_timer()

    print ("k", k, "-> Accuracy", score, "%", "| Time", int(toc - tic))


    scores.append(score)