def test_matrix_difference_with_accuracy_etc():
print "\n-- 'matrix_difference' (precision/recall/accuracy/cosine), 'max_binary_matrix' --"
X_true = np.array([[2, 0, 0],
[2, 0, 2],
[0, 1, 0],
[0, 0, 3],
[0, 0, 3],
[1, 0, 2],
[0, 3, 3]])
X_pred = np.array([[1, 1, 2],
[2, 1, 2],
[3, 4, 0],
[1, 1, 2],
[2, 1, 1],
[1, 2, 2],
[1, 2.99, 3]])
X_true_b = max_binary_matrix(X_true)
X_pred_b = max_binary_matrix(X_pred)
X_pred_b1 = max_binary_matrix(X_pred, threshold=0.1)
print "X_true:\n", X_true
print "X_pred:\n", X_pred
print "X_true binary:\n", X_true_b
print "X_pred binary:\n", X_pred_b
print "X_pred binary with threshold 0.1:\n", X_pred_b1
ind = list([])
# ind = list([0, 1])
# ind = list([1, 2, 3, 4, 5])
# ind = list([0, 2, 3, 4, 5, 6])
print "\nPrecision:\n", matrix_difference(X_true, X_pred, ind, vector=True, similarity='precision')
print "*** type:", type (matrix_difference(X_true, X_pred, ind, vector=True, similarity='precision'))
print "Recall:\n", matrix_difference(X_true, X_pred, ind, vector=True, similarity='recall')
print "Accuracy:\n", matrix_difference(X_true, X_pred, ind, vector=True, similarity='accuracy')
cosine_list = matrix_difference(X_true, X_pred, ind, vector=True, similarity='cosine')
print "Cosine:\n", cosine_list
print "Cosine sorted:\n", sorted(cosine_list, reverse=True)
print "\nPrecision:\n", matrix_difference(X_true, X_pred, ind, similarity='precision')
print "Recall:\n", matrix_difference(X_true, X_pred, ind, similarity='recall')
print "Accuracy:\n", matrix_difference(X_true, X_pred, ind)
print "Cosine:\n", matrix_difference(X_true, X_pred, ind, similarity='cosine')
def test_matrix_difference():
print "\n-- 'matrix_difference' (cosine/cosine_ratio/l2), 'to_centering_beliefs' --"
X0 = np.array([[2, 0, 0],
[2, 0, 2],
[0, 1, 0],
[0, 0, 3],
[0, 0, 3],
[1, 0, 2],
[0, 3, 3],
[0, 0, 0],
[0, 1, 0],
[0, 1, 0],
[9, 9, 9],
[9, 9, 9],
[100, 100, 100],])
X1 = np.array([[1, 1, 2],
[2, 1, 2],
[3, 4, 0],
[1, 1, 2],
[2, 1, 1],
[1, 2, 2],
[1, 2, 3],
[0, 0, 0],
[1, 0, 0],
[0, 2, 0],
[9, 9, 9],
[8, 9, 9],
[100, 100, 101],])
print "X0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine', vector=True)
print "cosine:\n", result
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine_ratio:\n", result
result = matrix_difference(X0, X1, similarity='l2', vector=True)
print "l2:\n", result
X0 = np.array([[ 1. , 0. , 0. ],
[ 0.30804075, 0.56206462, 0.12989463],
[ 0.32434628, 0.33782686, 0.33782686],
[ 0.30804075, 0.12989463, 0.56206462],
[ 0.14009173, 0.71981654, 0.14009173],
[ 0.32273419, 0.21860539, 0.45866042],
[ 0.33804084, 0.32391832, 0.33804084],
[ 0.45866042, 0.21860539, 0.32273419]])
X1 = np.array([[ 1. , 0. , 0. ],
[ 0.22382029, 0.45296374, 0.32321597],
[ 0.32434628, 0.33782686, 0.33782686],
[ 0.22382029, 0.32321597, 0.45296374],
[ 0.2466463 , 0.5067074 , 0.2466463 ],
[ 0.32273419, 0.21860539, 0.45866042],
[ 0.33804084, 0.32391832, 0.33804084],
[ 0.45866042, 0.21860539, 0.32273419]])
print "\nX0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine:\n", result
# X0z = row_normalize_matrix(X0, norm='zscores')
# X1z = row_normalize_matrix(X1, norm='zscores')
X0z = to_centering_beliefs(X0)
X1z = to_centering_beliefs(X1)
print "\nX0z:\n", X0z
print "X1z:\n", X1z
result = matrix_difference(X0z, X1z, similarity='cosine_ratio', vector=True)
print "cosine zscores:\n", result
# actualPercentageConverged = matrix_convergence_percentage(X0z, X1z, threshold=convergenceCosineSimilarity)
X0 = np.array([1, 0, 0])
X1 = np.array([1, 1, 0])
print "\nX0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine zscores:\n", result
def test_matrix_difference_with_cosine_simililarity():
print "\n-- 'matrix_difference' (cosine), 'row_normalize_matrix' --"
print "k=3"
v1 = np.array([1, 0, 0])
v2 = np.array([0, 1, 0])
v3 = np.array([1, 1, 0])
print "Cosine with original:\n ", \
matrix_difference(v1,
v1, similarity='cosine')
print "Cosine with original zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v1, norm='zscores'), similarity='cosine')
print "Cosine with zscore :\n ", \
matrix_difference(v1,
row_normalize_matrix(v1, norm='zscores'), similarity='cosine')
print "Cosine with normal:\n ", \
matrix_difference(v1,
v2, similarity='cosine')
print "Cosine with normal after both zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v2, norm='zscores'), similarity='cosine')
print "! Notice that average guessing leads to expectation of 0!"
print "Cosine v1, v3:\n ", \
matrix_difference(v1,
v3, similarity='cosine')
print "Cosine v1, v3 after zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "\nk=5"
v1 = np.array([1, 0, 0, 0, 0])
v2 = np.array([0, 1, 0, 0, 0])
v3 = np.array([1, 1, 0, 0, 0])
v4 = np.array([0, 0, 0, 0, 0])
print "Cosine with normal:\n ", \
matrix_difference(v1,
v2, similarity='cosine')
print "Cosine with normal after both zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v2, norm='zscores'), similarity='cosine')
print "! Notice that average guessing leads to expectation of 0!"
print "Cosine v1, v3:\n ", \
matrix_difference(v1,
v3, similarity='cosine')
print "Cosine v1, v3 after zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "Average Cos similarity partly zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "Cosine with 0-vector:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v4, norm='zscores'), similarity='cosine')
print
X = np.array([[1, 0, 0, 0, 0],
[1, 0, 0, 0, 0],
[1, 0, 0, 0, 0],
[1, 0, 0, 0, 0],
[1, 1, 0, 0, 0]])
Y = np.array([[1, 0, 0, 0, 0],
[1, 1, 0, 0, 0],
[0, 0, 0, 0, 0],
[0, 1, 0, 0, 0],
[1, 1.1, 0, 0, 0]])
print "X\n", X
print "Y\n", Y
Xs = row_normalize_matrix(X, norm='zscores')
Ys = row_normalize_matrix(Y, norm='zscores')
print "Xs\n", Xs
print "Ys\n", Ys
print "\nCosine original:\n ", \
matrix_difference(X,
Y, vector=True, similarity='cosine')
print "Cosine zscore:\n ", \
matrix_difference(Xs,
Ys, vector=True, similarity='cosine')
print "Average cosine zscore:\n ", \
matrix_difference(X,
Y, similarity='cosine')
def test_matrix_difference():
print "\n-- 'matrix_difference' (cosine/cosine_ratio/l2), 'to_centering_beliefs' --"
X0 = np.array([
[2, 0, 0],
[2, 0, 2],
[0, 1, 0],
[0, 0, 3],
[0, 0, 3],
[1, 0, 2],
[0, 3, 3],
[0, 0, 0],
[0, 1, 0],
[0, 1, 0],
[9, 9, 9],
[9, 9, 9],
[100, 100, 100],
])
X1 = np.array([
[1, 1, 2],
[2, 1, 2],
[3, 4, 0],
[1, 1, 2],
[2, 1, 1],
[1, 2, 2],
[1, 2, 3],
[0, 0, 0],
[1, 0, 0],
[0, 2, 0],
[9, 9, 9],
[8, 9, 9],
[100, 100, 101],
])
print "X0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine', vector=True)
print "cosine:\n", result
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine_ratio:\n", result
result = matrix_difference(X0, X1, similarity='l2', vector=True)
print "l2:\n", result
X0 = np.array([[1., 0., 0.], [0.30804075, 0.56206462, 0.12989463],
[0.32434628, 0.33782686, 0.33782686],
[0.30804075, 0.12989463, 0.56206462],
[0.14009173, 0.71981654, 0.14009173],
[0.32273419, 0.21860539, 0.45866042],
[0.33804084, 0.32391832, 0.33804084],
[0.45866042, 0.21860539, 0.32273419]])
X1 = np.array([[1., 0., 0.], [0.22382029, 0.45296374, 0.32321597],
[0.32434628, 0.33782686, 0.33782686],
[0.22382029, 0.32321597, 0.45296374],
[0.2466463, 0.5067074, 0.2466463],
[0.32273419, 0.21860539, 0.45866042],
[0.33804084, 0.32391832, 0.33804084],
[0.45866042, 0.21860539, 0.32273419]])
print "\nX0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine:\n", result
# X0z = row_normalize_matrix(X0, norm='zscores')
# X1z = row_normalize_matrix(X1, norm='zscores')
X0z = to_centering_beliefs(X0)
X1z = to_centering_beliefs(X1)
print "\nX0z:\n", X0z
print "X1z:\n", X1z
result = matrix_difference(X0z,
X1z,
similarity='cosine_ratio',
vector=True)
print "cosine zscores:\n", result
# actualPercentageConverged = matrix_convergence_percentage(X0z, X1z, threshold=convergenceCosineSimilarity)
X0 = np.array([1, 0, 0])
X1 = np.array([1, 1, 0])
print "\nX0:\n", X0
print "X1:\n", X1
result = matrix_difference(X0, X1, similarity='cosine_ratio', vector=True)
print "cosine zscores:\n", result
def test_matrix_difference_with_accuracy_etc():
print "\n-- 'matrix_difference' (precision/recall/accuracy/cosine), 'max_binary_matrix' --"
X_true = np.array([[2, 0, 0], [2, 0, 2], [0, 1, 0], [0, 0, 3], [0, 0, 3],
[1, 0, 2], [0, 3, 3]])
X_pred = np.array([[1, 1, 2], [2, 1, 2], [3, 4, 0], [1, 1, 2], [2, 1, 1],
[1, 2, 2], [1, 2.99, 3]])
X_true_b = max_binary_matrix(X_true)
X_pred_b = max_binary_matrix(X_pred)
X_pred_b1 = max_binary_matrix(X_pred, threshold=0.1)
print "X_true:\n", X_true
print "X_pred:\n", X_pred
print "X_true binary:\n", X_true_b
print "X_pred binary:\n", X_pred_b
print "X_pred binary with threshold 0.1:\n", X_pred_b1
ind = list([])
# ind = list([0, 1])
# ind = list([1, 2, 3, 4, 5])
# ind = list([0, 2, 3, 4, 5, 6])
print "\nPrecision:\n", matrix_difference(X_true,
X_pred,
ind,
vector=True,
similarity='precision')
print "*** type:", type(
matrix_difference(X_true,
X_pred,
ind,
vector=True,
similarity='precision'))
print "Recall:\n", matrix_difference(X_true,
X_pred,
ind,
vector=True,
similarity='recall')
print "Accuracy:\n", matrix_difference(X_true,
X_pred,
ind,
vector=True,
similarity='accuracy')
cosine_list = matrix_difference(X_true,
X_pred,
ind,
vector=True,
similarity='cosine')
print "Cosine:\n", cosine_list
print "Cosine sorted:\n", sorted(cosine_list, reverse=True)
print "\nPrecision:\n", matrix_difference(X_true,
X_pred,
ind,
similarity='precision')
print "Recall:\n", matrix_difference(X_true,
X_pred,
ind,
similarity='recall')
print "Accuracy:\n", matrix_difference(X_true, X_pred, ind)
print "Cosine:\n", matrix_difference(X_true,
X_pred,
ind,
similarity='cosine')
def test_matrix_difference_with_cosine_simililarity():
print "\n-- 'matrix_difference' (cosine), 'row_normalize_matrix' --"
print "k=3"
v1 = np.array([1, 0, 0])
v2 = np.array([0, 1, 0])
v3 = np.array([1, 1, 0])
print "Cosine with original:\n ", \
matrix_difference(v1,
v1, similarity='cosine')
print "Cosine with original zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v1, norm='zscores'), similarity='cosine')
print "Cosine with zscore :\n ", \
matrix_difference(v1,
row_normalize_matrix(v1, norm='zscores'), similarity='cosine')
print "Cosine with normal:\n ", \
matrix_difference(v1,
v2, similarity='cosine')
print "Cosine with normal after both zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v2, norm='zscores'), similarity='cosine')
print "! Notice that average guessing leads to expectation of 0!"
print "Cosine v1, v3:\n ", \
matrix_difference(v1,
v3, similarity='cosine')
print "Cosine v1, v3 after zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "\nk=5"
v1 = np.array([1, 0, 0, 0, 0])
v2 = np.array([0, 1, 0, 0, 0])
v3 = np.array([1, 1, 0, 0, 0])
v4 = np.array([0, 0, 0, 0, 0])
print "Cosine with normal:\n ", \
matrix_difference(v1,
v2, similarity='cosine')
print "Cosine with normal after both zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v2, norm='zscores'), similarity='cosine')
print "! Notice that average guessing leads to expectation of 0!"
print "Cosine v1, v3:\n ", \
matrix_difference(v1,
v3, similarity='cosine')
print "Cosine v1, v3 after zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "Average Cos similarity partly zscore:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v3, norm='zscores'), similarity='cosine')
print "Cosine with 0-vector:\n ", \
matrix_difference(row_normalize_matrix(v1, norm='zscores'),
row_normalize_matrix(v4, norm='zscores'), similarity='cosine')
print
X = np.array([[1, 0, 0, 0, 0], [1, 0, 0, 0, 0], [1, 0, 0, 0, 0],
[1, 0, 0, 0, 0], [1, 1, 0, 0, 0]])
Y = np.array([[1, 0, 0, 0, 0], [1, 1, 0, 0, 0], [0, 0, 0, 0, 0],
[0, 1, 0, 0, 0], [1, 1.1, 0, 0, 0]])
print "X\n", X
print "Y\n", Y
Xs = row_normalize_matrix(X, norm='zscores')
Ys = row_normalize_matrix(Y, norm='zscores')
print "Xs\n", Xs
print "Ys\n", Ys
print "\nCosine original:\n ", \
matrix_difference(X,
Y, vector=True, similarity='cosine')
print "Cosine zscore:\n ", \
matrix_difference(Xs,
Ys, vector=True, similarity='cosine')
print "Average cosine zscore:\n ", \
matrix_difference(X,
Y, similarity='cosine')
