def find_best_k(func, method, td_lag):
# Find embedding dimension which maximizes AP
k_best, ap_best, auc_best = 0, 0.0, 0.0
regions_best = []
for k in range(3, 21):
detections = maxdiv.maxdiv(func['ts'],
method=method,
mode='I_OMEGA',
extint_min_len=20,
extint_max_len=100,
num_intervals=None,
td_dim=k,
td_lag=td_lag)
cur_ap = eval.average_precision([func['gt']], [detections])
cur_auc = eval.auc(func['gt'], detections, func['ts'].shape[1])
if (k_best == 0) or (cur_ap > ap_best) or ((cur_ap == ap_best) and
(cur_auc > auc_best)):
k_best, ap_best, auc_best, regions_best = k, cur_ap, cur_auc, detections
return regions_best, k_best
ratios = []
for l in range(20, 201, 15):
print(l)
funcs = [sample_gp_with_meanshift(n, l) for i in range(m)]
for method in methods:
auc = []
regions = []
for i in range(m):
gp, ygt = funcs[i]
regions.append(maxdiv.maxdiv(gp, method = method, num_intervals = 5, extint_min_len = 20, extint_max_len = 220,
kernelparameters={'kernel_sigma_sq': args.kernel_sigma_sq}, **parameters))
auc.append(eval.auc(ygt, regions[-1], n))
aucs[method].append(np.mean(auc))
aps[method].append(eval.average_precision([ygt for gp, ygt in funcs], regions))
ratios.append(float(l) / n)
# Plot results
fig_auc = plt.figure()
sp_auc = fig_auc.add_subplot(111, xlabel = 'Length of anomaly / Length of time series', ylabel = 'AUC')
fig_ap = plt.figure()
sp_ap = fig_ap.add_subplot(111, xlabel = 'Length of anomaly / Length of time series', ylabel = 'Average Precision')
for method in methods:
sp_auc.plot(ratios, aucs[method], marker = 'x', label = method)
sp_ap.plot(ratios, aps[method], marker = 'x', label = method)
sp_auc.legend(bbox_to_anchor=(0., 1.02, 1., .102), loc=3, ncol=len(methods), mode="expand", borderaxespad=0.)
aucs[ftype] = []
best_k[ftype] = Counter()
for i, func in enumerate(data[ftype]):
func_ids.append('{}_{:03d}'.format(ftype, i))
ygts.append(func['gt'])
det, k_best = optimizers[args.optimizer](func, args.method,
args.td_lag)
# Divide scores by maximum score since their range differs widely depending on the dimensionality
if args.method not in ('gaussian_cov_ts', 'gaussian_ts'):
for r in range(len(det) - 1, -1, -1):
det[r] = (det[r][0], det[r][1], det[r][2] / det[0][2])
regions.append(det)
aucs[ftype].append(eval.auc(func['gt'], det, func['ts'].shape[1]))
best_k[ftype][k_best] += 1
print("Best k: {}".format(k_best))
aps[ftype] = eval.average_precision(ygts, regions)
print("AP: {}".format(aps[ftype]))
if args.plot:
plt.bar(
np.array(list(best_k[ftype].keys())) - 0.5,
list(best_k[ftype].values()), 1)
plt.title(ftype)
plt.show()
all_ids += func_ids
all_regions += regions
maxdiv.maxdiv(
func['ts'],
useLibMaxDiv=True,
method=method,
preproc=preproc,
mode=mode,
td_dim=args.td_dim if preproc is not None else 1,
kernelparameters={
'kernel_sigma_sq': args.kernel_sigma_sq
},
**parameters))
time_stop = time.time()
ygts.append(func['gt'])
aucs.append(
eval.auc(func['gt'], regions[-1], func['ts'].shape[1]))
if preproc is None:
if func['ts'].shape[1] not in times:
times[func['ts'].shape[1]] = {
m: []
for m in METHODS
}
times[func['ts'].shape[1]][method].append(time_stop -
time_start)
auc[id][ftype] = np.mean(aucs)
auc_sd[id][ftype] = np.std(aucs)
aps[id][ftype] = eval.average_precision(ygts, regions)
all_gt[id] += ygts
all_regions[id] += regions