We visualize here the data cluster for the Wall Street Data presented in the columns version



In [2]:

    
import h5py
import matplotlib.pyplot as plt
import sys
sys.path.append("../")
%matplotlib inline
from visualization.data_clustering import visualize_data_cluster_text_to_image_columns

First let's do it for the version with spaces

Mixed Receptive Fields



In [3]:

    
# First we load the file 
file_location = '../results_database/text_wall_street_columns.hdf5'
run_name = '/test'
f = h5py.File(file_location, 'r')



In [4]:

    
# Nexa parameters
Nspatial_clusters = 3
Ntime_clusters = 3
Nembedding = 3

parameters_string = '/' + str(Nspatial_clusters)
parameters_string += '-' + str(Ntime_clusters)
parameters_string += '-' + str(Nembedding)

nexa = f[run_name + parameters_string]
cluster_to_index = nexa['cluster_to_index']

Let's see the receptive fields first



In [5]:

    
matrix = np.zeros((10, 3))

for cluster in cluster_to_index:
    
    cluster_indexes = cluster_to_index[str(cluster)]

    for index in cluster_indexes:
        first_index = index // 3
        second_index = index % 3
        matrix[first_index, second_index] = cluster

fig = plt.figure(figsize=(16, 12))
ax = fig.add_subplot(111)
ax.imshow(matrix, origin='lower', interpolation='none')









    Out[5]:





<matplotlib.image.AxesImage at 0x7f6b940a9c88>



In [6]:

    
cluster = 0
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)



In [7]:

    
cluster = 1
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)



In [8]:

    
cluster = 2
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)

Now for independent receptive fields



In [9]:

    
# First we load the file 
file_location = '../results_database/text_wall_street_columns.hdf5'
run_name = '/indep'
f = h5py.File(file_location, 'r')



In [10]:

    
# Nexa parameters
Nspatial_clusters = 3
Ntime_clusters = 3
Nembedding = 3

parameters_string = '/' + str(Nspatial_clusters)
parameters_string += '-' + str(Ntime_clusters)
parameters_string += '-' + str(Nembedding)

nexa = f[run_name + parameters_string]
cluster_to_index = nexa['cluster_to_index']



In [11]:

    
matrix = np.zeros((10, 3))

for cluster in cluster_to_index:
    
    cluster_indexes = cluster_to_index[str(cluster)]

    for index in cluster_indexes:
        first_index = index // 3
        second_index = index % 3
        matrix[first_index, second_index] = cluster

fig = plt.figure(figsize=(16, 12))
ax = fig.add_subplot(111)
ax.imshow(matrix, origin='lower', interpolation='none')









    Out[11]:





<matplotlib.image.AxesImage at 0x7f6b7edf21d0>



In [12]:

    
cluster = 0
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)



In [13]:

    
cluster = 1
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)



In [14]:

    
cluster = 2
for data_center in range(Ntime_clusters):
    fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)

Now let's do it for the version without spaces

Mixed Receptive Fields



In [16]:

    
# First we load the file 
file_location = '../results_database/text_wall_street_columns_spaces.hdf5'
run_name = '/test'
f = h5py.File(file_location, 'r')



In [17]:

    
# Nexa parameters
Nspatial_clusters = 3
Ntime_clusters = 3
Nembedding = 3

parameters_string = '/' + str(Nspatial_clusters)
parameters_string += '-' + str(Ntime_clusters)
parameters_string += '-' + str(Nembedding)

nexa = f[run_name + parameters_string]
cluster_to_index = nexa['cluster_to_index']



In [18]:

    
matrix = np.zeros((10, 3))

for cluster in cluster_to_index:
    
    cluster_indexes = cluster_to_index[str(cluster)]

    for index in cluster_indexes:
        first_index = index // 3
        second_index = index % 3
        matrix[first_index, second_index] = cluster

fig = plt.figure(figsize=(16, 12))
ax = fig.add_subplot(111)
ax.imshow(matrix, origin='lower', interpolation='none')









    Out[18]:





<matplotlib.image.AxesImage at 0x7f6b7eebb6a0>



In [20]:

    
for cluster in range(3):
    for data_center in range(Ntime_clusters):
        fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)

Now for independent receptive fields



In [21]:

    
# First we load the file 
file_location = '../results_database/text_wall_street_columns_spaces.hdf5'
run_name = '/indep'
f = h5py.File(file_location, 'r')



In [22]:

    
# Nexa parameters
Nspatial_clusters = 3
Ntime_clusters = 3
Nembedding = 3

parameters_string = '/' + str(Nspatial_clusters)
parameters_string += '-' + str(Ntime_clusters)
parameters_string += '-' + str(Nembedding)

nexa = f[run_name + parameters_string]
cluster_to_index = nexa['cluster_to_index']



In [23]:

    
matrix = np.zeros((10, 3))

for cluster in cluster_to_index:
    
    cluster_indexes = cluster_to_index[str(cluster)]

    for index in cluster_indexes:
        first_index = index // 3
        second_index = index % 3
        matrix[first_index, second_index] = cluster

fig = plt.figure(figsize=(16, 12))
ax = fig.add_subplot(111)
ax.imshow(matrix, origin='lower', interpolation='none')









    Out[23]:





<matplotlib.image.AxesImage at 0x7f6b7ebe5b38>



In [24]:

    
for cluster in range(3):
    for data_center in range(Ntime_clusters):
        fig = visualize_data_cluster_text_to_image_columns(nexa, f, run_name,
                                                       cluster, data_center, colorbar=True)



In [ ]: