def SOM_clustering():
splits_count = 4
selectedCab_index = 1
selectedCab = []
for q in open('sampled.txt', 'r').read().split():
selectedCab.append(int(q))
for split in range(1,splits_count+1):
arr = np.array([])
print("Split number : %d" % (split))
for cabID in selectedCab:
arrcol = np.array([])
#print(cabID)
filename = "..//Split//" + str(cabID) + "_" + str(split) + "_.traj"
with open(filename) as f:
for line in f:
line = line.rstrip()
arrcol = np.concatenate((arrcol,np.array([int(line)])),axis = 0)
#print(arrcol)
if len(arr) == 0:
arr = np.concatenate((arr,arrcol),axis = 0)
else:
arr = np.vstack((arr,arrcol))
testX = arr
print(testX)
print ("Cluster Start")
length = len(arr[0])
print("Dim : %d" %(length))
#Train a 20x30 SOM with 400 iterations
som = SOM(20, 30, length, 1)
som.train(testX)
print ("get_centroids")
#Get output grid
image_grid = som.get_centroids()
#Map colours to their closest neurons
mapped = som.map_vects(testX)
print ("Plot")
array = np.zeros((20,30))
filename_ClusterLabels = "Cabs_ClusterLabels" + str(split) +".csv"
target = open(filename_ClusterLabels,'w')
for i, m in enumerate(mapped):
index = ((m[0]) * 2) + (m[1]+1)
target.write("%s\n" % (index))
#print(index)
array[m[0], m[1]] += 1
target.close()
array = array.astype(int)
array = np.reshape(array,600)
print(array)
#break;
return;
formatter.generate_validation_file()
# Train SOM, if there is no weights.
if Path("weights.txt").is_file() is False:
# Use pandas for loading data using dataframes.
d = os.path.dirname(os.getcwd())
file_name = "huge_merged_csv_file.csv"
data = pd.read_csv(file_name, header=None)
# Shuffle the data in place.
data = data.sample(frac=1).reset_index(drop=True)
# create SOM object
som = SOM(7, 1)
# Train with 100 neurons.
som.train(data)
# Get output grid
grid = som.get_centroids()
# Save the weights in a file.
result_file = open("weights.txt", "w")
result_file.writelines(str(grid))
result_file.close()
# Map data to neurons.
mapped_vectors = mapper.map_vects()
tester.open_test_file()
tester.test_som(mapped_vectors)
def SOM_clustering():
arr = np.array([])
with open("AllTrajecsNopePoints.data") as f:
for line in f:
strings = line.split(" ")
arrcol = np.array([])
for str in strings:
arrcol = np.concatenate((arrcol, np.array([float(str)])),
axis=0)
if len(arr) == 0:
arr = np.concatenate((arr, arrcol), axis=0)
else:
arr = np.vstack((arr, arrcol))
#print (arr)
testX = arr
print("Cluster Start")
#Train a 20x30 SOM with 400 iterations
som = SOM(20, 30, 720, 400)
som.train(testX)
print("get_centroids")
#Get output grid
image_grid = som.get_centroids()
#Write to file
target = open("ClusteredCentroids.csv", 'w')
count = 0
for i, m in enumerate(image_grid):
for column in m:
for k, dim in enumerate(column):
if (k != 0):
target.write(" ")
target.write("%s" % (dim))
target.write("\n")
# target.write("%s " % (n))
# target.write("\n")
target.close()
#for i, m in enumerate(image_grid):
# print("i m[1] m[0] =", i ,m[1], m[0])
#Map colours to their closest neurons
mapped = som.map_vects(testX)
print("Plot")
array = np.zeros((20, 30))
#Plot
#plt.imshow(image_grid)
#plt.title('Color SOM')
#plt.plot(20, 30, 'b.')
target = open("ClusteredCabs_labels.csv", 'w')
for i, m in enumerate(mapped):
index = ((m[0]) * 30) + (m[1] + 1)
target.write("%s\n" % (index))
print(index)
array[m[0], m[1]] += 1
target.close()
#print("i m[1] m[0] =", i ,m[1], m[0])
#plt.plot(m[1], m[0], 'ro')
#plt.text(m[1], m[0], index, ha='center', va='center',
# bbox=dict(facecolor='white', alpha=0.5, lw=0))
#plt.show()
array = array.astype(int)
array = np.reshape(array, 600)
print(array)
#print (testX)
###Write to file
##target = open("ClusteredCabs.clu",'w')
##for i, m in enumerate(array):
## #if m != 0 :
## target.write("%s %s" % (i, m))
## target.write("\n")
##target.close()
return
for row in reader:
audio_features.append(list(row[i] for i in range(7, 17)))
track_names.append(list(row[i] for i in range(6, 7)))
i += 1
if i == num_tracks:
break
# Training inputs for tracks
audio_features = np.asarray(audio_features, dtype=float)
# Train a 20x30 SOM with 10 iterations
som = SOM(20, 30, 10, 50)
som.train(audio_features)
# Get output grid
image_grid = som.get_centroids()
# Map tracks to their closest neurons
mapped = som.map_vects(audio_features)
# Plot
plt.imshow(image_grid[10])
plt.title('Track SOM')
plt.ylim([0, 20])
plt.xlim([0, 30])
for i, m in enumerate(mapped):
plt.text(m[1],
m[0],
track_names[i],
ha='center',
va='center',
[0.125, 0.529, 1.0], [0.33, 0.4, 0.67], [0.6, 0.5, 1.0],
[0., 1., 0.], [1., 0., 0.], [0., 1., 1.], [1., 0., 1.],
[1., 1., 0.], [1., 1., 1.], [.33, .33, .33], [.5, .5, .5],
[.66, .66, .66]])
colorNames = [
'black', 'blue', 'darkblue', 'skyblue', 'greyblue', 'lilac', 'green',
'red', 'cyan', 'violet', 'yellow', 'white', 'darkgrey', 'mediumgrey',
'lightgrey'
]
#Train a 20x30 SOM with 400 iterations
som = SOM(m=20, n=30, dim=3, epochs=400)
som.train(colors)
#Get output grid
imageGrid = som.get_centroids()
#Map colours to their closest neurons
mapped = som.map_vects(colors)
#Plot
plt.imshow(imageGrid)
plt.title('Color SOM')
for i, m in enumerate(mapped):
plt.text(m[1],
m[0],
colorNames[i],
ha='center',
va='center',
bbox=dict(facecolor='white', alpha=0.5, lw=0))
plt.show()
