def nearest_center(centers, point):
nearest = centers[0]
for center in centers:
if dist(center, point) < dist(nearest, point):
nearest = center
return centers.index( nearest )
def nearest_center(centers, point):
nearest = centers[0]
for center in centers:
if dist(center, point) < dist(nearest, point):
nearest = center
return centers.index(nearest)
def sort_segments(segments): for i in xrange(len(segments)): for j in xrange(i,len(segments)): if dist(segments[i][2:],segments[i][:2]) >= dist(segments[j][2:],segments[j][:2]): tmp = segments[j] segments[j] = segments[i] segments[i] = tmp return
def sort_segments(segments):
for i in xrange(len(segments)):
for j in xrange(i, len(segments)):
if dist(segments[i][2:], segments[i][:2]) >= dist(
segments[j][2:], segments[j][:2]):
tmp = segments[j]
segments[j] = segments[i]
segments[i] = tmp
return
def equilateral_triangle(side):
pixs = []
cols = []
t1 = [-side / 2, 0]
t2 = [side / 2, 0]
t3 = [0, (sqrt(3) * side) / 2]
for x in xrange(t1[0], t2[0]):
for y in xrange(int(t3[1])):
alt = t3[1]
if y <= sqrt(3) * x + alt and y <= -sqrt(3) * x + alt:
# minus y-coordinate - [0,0] is top left and pos. directions are v and >
pixs.append([x + side, -y + side])
col = tuple(map(lambda x: int(2.3 * x), [dist([x, y], t1), dist([x, y], t2), dist([x, y], t3)]))
cols.append(col)
pixs = pixs_and_cols(pixs, cols)
return pixs
def equilateral_triangle(side):
pixs = []
cols = []
t1 = [-side / 2, 0]
t2 = [ side / 2, 0]
t3 = [ 0, (sqrt(3) * side) / 2]
for x in xrange(t1[0], t2[0]):
for y in xrange(int( t3[1] )):
alt = t3[1]
if y <= sqrt(3) * x + alt and y <= -sqrt(3) * x + alt:
# minus y-coordinate - [0,0] is top left and pos. directions are v and >
pixs.append([x + side,-y + side])
col = tuple( map( lambda x: int(2.3 * x), [ dist([x, y], t1),\
dist([x, y], t2),\
dist([x, y], t3),]))
cols.append( col )
pixs = pixs_and_cols(pixs, cols)
return pixs
def k_means(data, k, sorted_centers=False):
data = lists_to_pairs(data)
if sorted_centers:
data = [[x[0], x[1], dist([0, 0], [x[0], x[1]])] for x in data]
data.sort(key=operator.itemgetter(2))
n = len(data) - 1
tmp_data_A = data[::]
centers = []
for i in xrange(k):
tmp_data_B = tmp_data_A[:n / k]
centers.append(random.choice(tmp_data_B))
tmp_data_A = tmp_data_A[n / k:]
else:
anchor = find_center(data)
dif = max(min_max_points(data)[2:]) / k**2
centers = []
for i in xrange(k):
centers.append(
list(anchor + array([uniform(-dif, dif),
uniform(-dif, dif)])))
classes = [[] for i in xrange(k)]
tmp_class = []
all_centers = list(centers)
iteration = 0
while tmp_class != classes:
tmp_class = list(classes)
classes = [[] for i in xrange(k)]
for point in data:
classes[nearest_center(centers, point)].append(point)
for i, center in enumerate(centers):
centers[i] = find_center(classes[i], center)
iteration += 1
all_centers.extend(centers)
return classes, all_centers
def plot_shifts(centers, k, colors, filename=''):
iteration = len( centers ) / k - 1
for i in xrange(k):
tmp_centers = centers[i::k]
values = []
for j in xrange( iteration ):
values.append( dist(tmp_centers[j], tmp_centers[j + 1]) )
plt.plot(range(iteration), values, '-s', color=colors[i])
plt.title('Differences of centroids')
plt.xlabel('iteration $'+ str(iteration) +'$')
plt.ylabel('$\Delta C$')
if filename:
plt.savefig('img/'+ filename +'_shifts.png')
else:
plt.show()
plt.clf()
return
def plot_shifts(centers, k, colors, filename=''):
iteration = len(centers) / k - 1
for i in xrange(k):
tmp_centers = centers[i::k]
values = []
for j in xrange(iteration):
values.append(dist(tmp_centers[j], tmp_centers[j + 1]))
plt.plot(range(iteration), values, '-s', color=colors[i])
plt.title('Differences of centroids')
plt.xlabel('iteration $' + str(iteration) + '$')
plt.ylabel('$\Delta C$')
if filename:
plt.savefig('img/' + filename + '_shifts.png')
else:
plt.show()
plt.clf()
return
def k_means(data, k, sorted_centers=False):
data = lists_to_pairs( data )
if sorted_centers:
data = [[x[0], x[1], dist([0,0], [x[0],x[1]])] for x in data]
data.sort(key=operator.itemgetter(2))
n = len( data ) - 1
tmp_data_A = data[::]
centers = []
for i in xrange(k):
tmp_data_B = tmp_data_A[:n / k]
centers.append( random.choice( tmp_data_B ) )
tmp_data_A = tmp_data_A[n / k:]
else:
anchor = find_center( data )
dif = max( min_max_points( data )[2:]) / k**2
centers = []
for i in xrange(k):
centers.append(list(anchor + array([uniform(-dif, dif), uniform(-dif, dif)])))
classes = [ [] for i in xrange(k) ]
tmp_class = []
all_centers = list(centers)
iteration = 0
while tmp_class != classes:
tmp_class = list(classes)
classes = [ [] for i in xrange(k) ]
for point in data:
classes[ nearest_center(centers, point) ].append( point )
for i, center in enumerate(centers):
centers[ i ] = find_center( classes[ i ], center )
iteration += 1
all_centers.extend( centers )
return classes, all_centers
