In [1]:
import networkx
import math
import scipy.optimize
import numpy
from lib.time_graph import *
from lib.time_graph_signal_proc import *
from lib.vis import *
from lib.syn import *
from lib.io import *
from lib.datasets import *
from lib.time_graph import *
from lib.experiments import *
from IPython.display import Image
In [2]:
from datetime import datetime, date, time, timedelta
import statistics
def check_time_range_traffic(file_id, days, hours):
start_time = datetime.strptime("1/04/11 00:00", "%d/%m/%y %H:%M")
file_time = start_time + timedelta(minutes=int(file_id)*5)
check_day = False
for d in days:
if file_time.weekday() == d:
check_day = True
break
if check_day is False:
return False
check_hour = False
for h in hours:
if file_time.hour == h:
check_hour = True
break
return check_hour
In [3]:
friday_snaps = []
days = [4]
hours=range(24)
for i in range(0, 8640):
if check_time_range_traffic(i, days, hours):
friday_snaps.append(i)
friday_snaps = numpy.array(friday_snaps)
friday_snaps = friday_snaps[12*12:12*24]
friday_snaps = friday_snaps[0::36]
In [4]:
sG = read_graph(small_traffic["path"] + "traffic.graph", small_traffic["path"] + "traffic.data")
In [5]:
G = TimeGraph(0.)
for i in range(len(friday_snaps)):
for e in sG.edges():
G.add_edge(e[0], e[1], i)
In [6]:
F = read_time_values(small_traffic["path"]+"traffic", G, friday_snaps)
In [7]:
tc = TemporalCuts("name", "diff-sparse")
c = tc.cut(G)
print(c)
{'cut': array([ 1., 1., 1., 1., 1., -1., -1., -1., -1., -1., -1., -1., -1.,
1., 1., 1., 1., 1., -1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., -1., -1., -1., -1., -1., 1., 1., 1., 1.,
1., -1., -1., -1., -1., 1., 1., 1., 1., 1., 1., 1., -1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., -1., -1., -1., -1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., -1., -1., -1., -1., -1.,
1., 1., 1., 1., 1., 1., 1., -1., -1., 1., 1., 1., 1.,
1., -1., -1., -1., -1., -1., -1., -1., -1., 1., 1., 1., 1.,
1., -1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
-1., -1., -1., -1., -1., 1., 1., 1., 1., 1., -1., -1., -1.,
-1., 1., 1., 1., 1., 1., 1., 1., -1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., -1., -1.,
-1., -1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., -1., -1., -1., -1., -1., 1., 1., 1., 1.,
1., 1., 1., -1., -1., 1., 1., 1., 1., 1., -1., -1., -1.,
-1., -1., -1., -1., -1., 1., 1., 1., 1., 1., -1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., -1., -1., -1., -1.,
-1., 1., 1., 1., 1., 1., -1., -1., -1., -1., 1., 1., 1.,
1., 1., 1., 1., -1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., -1., -1., -1., -1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
-1., -1., -1., -1., -1., 1., 1., 1., 1., 1., 1., 1., -1.,
-1., 1., 1., 1., 1., 1., -1., -1., -1., -1., -1., -1., -1.,
-1., 1., 1., 1., 1., 1., -1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., -1., -1., -1., -1., -1., 1., 1., 1.,
1., 1., -1., -1., -1., -1., 1., 1., 1., 1., 1., 1., 1.,
-1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., -1., -1., -1., -1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., -1., -1., -1., -1.,
-1., 1., 1., 1., 1., 1., 1., 1., -1., -1.]), 'score': 0.0004761904761904762, 'edges': 4.0, 'swaps': 0.0}
In [8]:
draw_time_graph_cut(G, "primary_school_td.svg", c["cut"])
/usr/local/lib/python3.6/dist-packages/networkx/drawing/nx_pylab.py:611: MatplotlibDeprecationWarning: isinstance(..., numbers.Number) if cb.is_numlike(alpha):
In [8]:
c = temporal_graph_transform(G, F, 2, 0.)
#c = temporal_graph_transform(G, numpy.ones(Fshape[0]), .0001)
In [9]:
draw_time_graph_cut_values(G, c["assign"], F, "primary_school_td.svg")
/usr/local/lib/python3.6/dist-packages/networkx/drawing/nx_pylab.py:611: MatplotlibDeprecationWarning: isinstance(..., numbers.Number) if cb.is_numlike(alpha):
Out[9]:
{'0': array([ 0.18250847, -0.15794278]),
'86': array([ 0.30310487, -0.20965114]),
'32': array([ 0.06987987, -0.0838195 ]),
'25': array([ 0.4225651 , -0.23740256]),
'15': array([-0.03598401, 0.01582715]),
'1': array([-0.67924587, 0.32749639]),
'12': array([-0.75949408, 0.37252969]),
'38': array([-0.63020302, 0.35149615]),
'58': array([-0.5970037 , 0.27466767]),
'87': array([-0.82077516, 0.42929772]),
'17': array([-0.70000208, 0.43649227]),
'48': array([-0.54186503, 0.26780126]),
'22': array([-0.52596056, 0.23192355]),
'2': array([-0.2243784 , 0.13293801]),
'72': array([-0.32404565, 0.20205026]),
'62': array([-0.12201514, 0.03567913]),
'90': array([-0.18259888, 0.1502331 ]),
'7': array([-0.25037719, 0.21747988]),
'36': array([-0.44670351, 0.22944968]),
'43': array([-0.00229913, 0.10767679]),
'81': array([-0.16954768, -0.09943115]),
'68': array([-0.10839213, 0.05334948]),
'3': array([-0.24898908, -0.80134115]),
'84': array([-0.30572616, -0.83581881]),
'49': array([-0.20588472, -0.88959924]),
'14': array([-0.28161772, -0.69364204]),
'4': array([-0.33353714, -0.90994779]),
'21': array([-0.31168058, -0.74988836]),
'28': array([-0.18775438, -0.96405745]),
'53': array([-0.30919261, -0.5669643 ]),
'5': array([-0.91628616, 0.54044631]),
'89': array([-0.9373539 , 0.58321715]),
'34': array([-0.86983564, 0.49531281]),
'69': array([-0.86241501, 0.54832194]),
'79': array([-1. , 0.61287164]),
'6': array([-0.09403134, 0.14770653]),
'70': array([-0.04916571, 0.0920583 ]),
'55': array([-0.04488249, 0.1302277 ]),
'73': array([0.06722174, 0.0752563 ]),
'51': array([-0.01931539, 0.19806565]),
'8': array([0.39341628, 0.5494325 ]),
'97': array([0.35205617, 0.47532322]),
'67': array([0.299742 , 0.44779107]),
'91': array([0.31050551, 0.38817362]),
'9': array([0.76207623, 0.20153364]),
'23': array([0.72224938, 0.22783425]),
'99': array([0.81929201, 0.24897942]),
'45': array([0.69356972, 0.14808059]),
'11': array([0.79191568, 0.29209617]),
'42': array([0.63776714, 0.15868927]),
'24': array([0.60806907, 0.09556875]),
'10': array([-0.7782612 , 0.50493499]),
'66': array([0.84667124, 0.3480361 ]),
'13': array([0.40555411, 0.05765848]),
'16': array([0.29922913, 0.05629079]),
'18': array([0.51337064, 0.06452052]),
'44': array([0.40988923, 0.0340194 ]),
'30': array([0.19648692, 0.08110081]),
'65': array([0.28887517, 0.01837441]),
'31': array([0.52556496, 0.09309966]),
'46': array([-0.30739898, -0.44271557]),
'20': array([-0.31633865, -0.66615772]),
'85': array([-0.18030854, -0.07099775]),
'19': array([0.1738795 , 0.03667884]),
'60': array([0.10659391, 0.1377561 ]),
'71': array([-0.28791866, -0.5762822 ]),
'94': array([-0.26839832, -0.46705179]),
'98': array([-0.49156488, 0.18242478]),
'40': array([-0.56610687, 0.14669971]),
'95': array([-0.36782521, 0.14225357]),
'83': array([-0.42534436, 0.17265405]),
'56': array([ 0.53529357, -0.24233811]),
'27': array([ 0.60131454, -0.31310825]),
'52': array([ 0.57258205, -0.1638762 ]),
'26': array([-0.2341893, -0.1942139]),
'37': array([-0.21300749, -0.22632152]),
'33': array([-0.27702058, -0.31858063]),
'47': array([-0.24537887, -0.34643236]),
'39': array([ 0.60301813, -0.40214144]),
'74': array([ 0.60485702, -0.26165828]),
'29': array([ 0.50662661, -0.40174664]),
'88': array([ 0.66870426, -0.46005252]),
'50': array([ 0.57136335, -0.34435066]),
'78': array([ 0.3894977 , -0.43449696]),
'80': array([ 0.27711547, -0.45692074]),
'76': array([0.06253763, 0.23321054]),
'35': array([0.19725101, 0.22547405]),
'61': array([0.26974763, 0.3014139 ]),
'63': array([0.09328786, 0.18542861]),
'93': array([0.17460215, 0.27755079]),
'82': array([-0.0573306 , 0.12219849]),
'41': array([ 0.1817284 , -0.47988771]),
'64': array([ 0.1172176 , -0.51001667]),
'96': array([0.16284315, 0.1393929 ]),
'57': array([0.00482533, 0.27491996]),
'75': array([0.31537127, 0.13355094]),
'92': array([ 0.52789467, -0.03963841]),
'77': array([0.44309983, 0.07362373]),
'54': array([-0.21919693, 0.11622148]),
'59': array([0.2513154 , 0.36562967])}
In [10]:
draw_time_graph_partitions(G, "primary_school_td.svg", c["assign"], False)
In [11]:
L2_error(F, c['transform'])
Out[11]:
0.19050809628051743
In [8]:
K = [5, 10, 20, 40]
results = compression_experiments(G, F, K, 5)
In [9]:
results
Out[9]:
array([[ 0.87305588, 0.70768239, 0.59836889, 0.46158191],
[ 0.19189918, 0.17817595, 0.17081546, 0.15817729],
[ 0.22343697, 0.21890282, 0.20816496, 0.18720717],
[ 0.28669265, 0.27703713, 0.25879688, 0.23989775]])
In [2]:
K = [5, 10, 20, 40]
results = numpy.array([[ 0.87305588, 0.70768239, 0.59836889, 0.46158191],
[ 0.1925559 , 0.17970188, 0.16977256, 0.16329276],
[ 0.23102427, 0.2132812 , 0.19909818, 0.18990198],
[ 0.28271556, 0.27284988, 0.26597022, 0.26199616]])
In [10]:
output_file_name = "figs/compression_traffic.png"
plot_compression_experiments(results, K, output_file_name, .9, 9, 0.1, 1.)
Image(filename=output_file_name)
Out[10]:
In [12]:
G = read_time_graph(primary_school["graph"], 0.05)
F = generate_dyn_heat(G, 1, 15.)
F = F.flatten()
F = F - numpy.mean(F)
F = F / numpy.linalg.norm(F)
In [13]:
K = [5, 10, 20, 40]
results = compression_experiments(G, F, K, 5)
In [20]:
output_file_name = "figs/compression_heat.png"
plot_compression_experiments(results, K, output_file_name, .9, 9, 0.05,1.)
Image(filename=output_file_name)
Out[20]:
