Permalink
Fetching contributors…
Cannot retrieve contributors at this time
123 lines (107 sloc) 3.46 KB
# Copyright 2004-2008 by Michiel de Hoon. All rights reserved.
# This code is part of the Biopython distribution and governed by its
# license. Please see the LICENSE file that should have been included
# as part of this package.
# See the Biopython Tutorial for an explanation of the biological
# background of these tests.
import unittest
try:
import numpy
del numpy
from numpy import asarray
del asarray
except ImportError:
from Bio import MissingPythonDependencyError
raise MissingPythonDependencyError(
"Install NumPy if you want to use Bio.kNN.")
from Bio import kNN
xs = [[-53, -200.78],
[117, -267.14],
[57, -163.47],
[16, -190.30],
[11, -220.94],
[85, -193.94],
[16, -182.71],
[15, -180.41],
[-26, -181.73],
[58, -259.87],
[126, -414.53],
[191, -249.57],
[113, -265.28],
[145, -312.99],
[154, -213.83],
[147, -380.85],
[93, -291.13]]
ys = [1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
0,
0,
0,
0,
0,
0,
0]
class TestKNN(unittest.TestCase):
def test_calculate_model(self):
k = 3
model = kNN.train(xs, ys, k)
self.assertEqual(model.classes, set([0, 1]))
n = len(xs)
for i in range(n):
self.assertAlmostEqual(model.xs[i, 0], xs[i][0], places=4)
self.assertAlmostEqual(model.xs[i, 1], xs[i][1], places=4)
self.assertEqual(model.ys[i], ys[i])
self.assertEqual(model.k, k)
def test_classify(self):
k = 3
model = kNN.train(xs, ys, k)
result = kNN.classify(model, [6, -173.143442352])
self.assertEqual(result, 1)
result = kNN.classify(model, [309, -271.005880394])
self.assertEqual(result, 0)
def test_calculate_probability(self):
k = 3
model = kNN.train(xs, ys, k)
weights = kNN.calculate(model, [6, -173.143442352])
self.assertAlmostEqual(weights[0], 0.0, places=6)
self.assertAlmostEqual(weights[1], 3.0, places=6)
weights = kNN.calculate(model, [309, -271.005880394])
self.assertAlmostEqual(weights[0], 3.0, places=6)
self.assertAlmostEqual(weights[1], 0.0, places=6)
weights = kNN.calculate(model, [117, -267.13999999999999])
self.assertAlmostEqual(weights[0], 2.0, places=6)
self.assertAlmostEqual(weights[1], 1.0, places=6)
def test_model_accuracy(self):
correct = 0
k = 3
model = kNN.train(xs, ys, k)
predictions = [1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0]
for i in range(len(predictions)):
prediction = kNN.classify(model, xs[i])
self.assertEqual(prediction, predictions[i])
if prediction == ys[i]:
correct += 1
self.assertEqual(correct, 15)
def test_leave_one_out(self):
correct = 0
k = 3
model = kNN.train(xs, ys, k)
predictions = [1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1]
for i in range(len(predictions)):
model = kNN.train(xs[:i] + xs[i + 1:], ys[:i] + ys[i + 1:], k)
prediction = kNN.classify(model, xs[i])
self.assertEqual(prediction, predictions[i])
if prediction == ys[i]:
correct += 1
self.assertEqual(correct, 13)
if __name__ == "__main__":
runner = unittest.TextTestRunner(verbosity=2)
unittest.main(testRunner=runner)