def init_files(trace_type = ' ', task_type='CLASSIFICATION', topo_type='b4',
x_var = 'FLOWNUM', p=0.01
):
file_name1, file_name2 = cf.init_files(trace_type=trace_type,
task_type=task_type, topo_type=topo_type,
x_var=x_var, p=p)
return file_name1, file_name2
def init_files(trace_type=' ',
task_type='CLASSIFICATION',
topo_type='b4',
x_var='FLOWNUM',
p=0.01):
file_name1, file_name2 = cf.init_files(trace_type=trace_type,
task_type=task_type,
topo_type=topo_type,
x_var=x_var,
p=p)
f1 = open(file_name1, 'w')
if task_type == 'APPLICATION':
f1.write((
"#flows,iteration,r_threshold,imr_threshold,p_t,topk,is_correlated,#case_detect,len(class_dict)\n"
))
else:
if task_type == 'CLASSIFICATION':
f1.write((
"#flows,iteration,epsilon,r_threshold,is_correlated,#correlated_detect,#non-uniform_detect,#all_cases\n"
))
elif task_type == 'RANKING':
f1.write((
"#flows,iteration,epsilon,r_threshold,is_correlated,#correlated_detect\n"
))
else:
f1.write((
"#flows,iteration,r_threshold,imr_threshold,is_correlated,#correlated_detect,#non-uniform_detect,#all_cases\n"
))
f2 = open(file_name2, 'w')
if task_type == 'CLASSIFICATION':
f2.write('#flows,iteration,time,epsilon,r_threshold,' +
'is_correlated,' + 'key,count_chern,byte_chern,pred,true\n')
elif task_type == 'RANKING':
f2.write('#flows,iteration,time,epsilon,r_threshold,' +
'is_correlated,' + 'key,#pres,o_f,count_chern[0]\n')
elif task_type == 'APPLICATION':
f2.write('#flows,iteration,time,r_threshold,imr_threshold,p_t,topk,' +
'is_correlated,' + 'drop_id,key,pred,true,imr0,imr1,imr2\n')
print(file_name1, '\n', file_name2)
f1.close()
f2.close()
def init_files(trace_type=' ',
task_type='CLASSIFICATION',
topo_type='b4',
x_var='FLOWNUM',
p=0.01):
"""file_name1 = ('../outputs/zout_'
+topo_type.lower()+'_'+ task_type.lower() + '_'
+ x_var.lower() + '_trace'+trace_type+'.csv')
file_name2 = ('../outputs/zout_'
+topo_type.lower()+'_'+ task_type.lower() + '_'
+ x_var.lower() + '_chern_trace'+trace_type+'.csv')"""
file_name1, file_name2 = cf.init_files(trace_type=trace_type,
task_type=task_type,
topo_type=topo_type,
x_var=x_var,
p=p)
return file_name1, file_name2
black_hole=black_hole,
links=links,
metric_type=x_var,
exc_time=exc_time,
app_exc_time=app_exc_time,
flow_paths=flow_paths)
print('trues, ests, select_len, est_pairs, exc_time, app_exc_time',
len(trues), len(ests), min_len, est_pairs, exc_time, app_exc_time)
classification_alg(true_byte_dict, epsilon=0.005, epsilon_pb=epsilon)
# Write to fsile
select_len = min_len
file_name1, file_name2 = cf.init_files(trace_type=trace_type,
task_type=task_type,
topo_type=topo_type,
x_var=x_var,
p=0.01)
f1 = open(file_name1, 'a')
f2 = open(file_name2, 'a')
if task_type == "CLASSIFICATION":
for key in true_byte_dict:
true_class = is_correlated
if drop_id <= 0:
true_class = -1
is_correlated = -1
f2.write(
str(select_len) + ',' + str(exc_time) + ',' + str(epsilon) +
',' + str(r_threshold) + ',' + str(int(is_correlated)) + ',' +
key + ',' + str(true_byte_dict[key][0]) + ',' +
def app_classification(app_flows1,
app_flows2,
app_flows3,
x_var='',
key='',
iteration=0,
is_correlated=0,
imr_threshold=0.1,
select_len=1,
app_exc_time=0,
r_threshold=0.1,
x_type=1,
p_t=0.01,
topk=1):
app1 = sorted(
Counter([randint(0, 8) for p in range(0, app_flows1)]).values())
app2 = sorted(
Counter([randint(0, 8) for p in range(0, app_flows2)]).values())
app3 = sorted(
Counter([randint(0, 8) for p in range(0, app_flows3)]).values())
#print 'r_threshold', r_threshold
drop_id = 3
if is_correlated == 0 and x_type == 2:
drop_link_id = randint(0, 8)
app_x = app3
app_x[drop_link_id] = int(app_x[drop_link_id] * (1 - r_threshold))
app_x = sorted(app_x)
app3 = app_x
elif is_correlated == 1 and x_type == 1:
for i in range(4):
app1[i], app2[i], app3[i] = 0.1, 0.1, 0.1
topk = topk
app_top1 = app1[0:topk] + app1[-topk:]
app_top2 = app2[0:topk] + app2[-topk:]
app_top3 = app3[0:topk] + app3[-topk:]
app_link_table = []
app_link_table.append(app_top1)
app_link_table.append(app_top2)
app_link_table.append(app_top3)
app_link_table = np.transpose(np.array(app_link_table))
top_apps = [1, 2, 3]
app_df = pd.DataFrame(app_link_table,
columns=['a' + str(x) for x in top_apps])
app_imrs = iset.independentCluster(app_df, p_t)
class_dict = {}
appl.appClassification(len(app_imrs),
app_imrs[0][2],
key=key,
imr_threshold=0.1,
class_dict=class_dict)
imrs = [x[2] for x in app_imrs]
file_name1, file_name2 = cf.init_files(trace_type='ABC',
task_type='APPLICATION',
topo_type='JUPITER',
x_var=x_var,
p=0.01)
f1 = open(file_name1, 'a')
f2 = open(file_name2, 'a')
if task_type == "APPLICATION":
# Ranking result
true_num = 0
for key in class_dict:
true_class = 0
true_class = x_type
f2.write(
str(select_len) + ',' + str(iteration) + ',' +
str(app_exc_time) + ',' + str(r_threshold) + ',' +
str(imr_threshold) + ',' + str(p_t) + ',' + str(topk) + ',' +
str(int(is_correlated)) + ',' + str(int(drop_id)) + ',' + key +
',' + str(class_dict[key]) + ',' + str(true_class) + ',' +
','.join([str(x) for x in imrs]) + '\n')
if true_class == class_dict[key]:
true_num += 1
f1.write(','.join([
str(x) for x in [
select_len, iteration, r_threshold, imr_threshold, p_t, topk,
int(is_correlated), true_num,
len(class_dict)
]
]) + '\n')