def plot_homogeneity_and_completeness2(category):
print(category)
D = DataGetter()
data = pandas.read_json(
D.fmt_path('datadir/homogeneity_and_completeness/' + category +
'.json'),
lines=True)
data = data.sort_values(by=['threshold'])
with open(
D.fmt_path('datadir/homogeneity_and_completeness/labelsets/' +
category + '.json'), 'r') as f:
labelset = json.load(f)
x = data.threshold.to_list()
y1 = data.homogeneity.to_list()
y2 = data.completeness.to_list()
y3 = data.nclusters.to_list()
fig, (ax) = plt.subplots(1, 1)
ax2 = ax.twinx()
l1 = ax.plot(x, y1, '--', label='homogeneity')
l2 = ax.plot(x, y2, label='completeness')
l3 = ax2.plot(x, y3, 'r:', label='# clusters')
ax2.axhline(len(labelset), color='k')
#vind, _ = min([(i,abs(n-len(labelset))) for i,n in enumerate(y3)], key=lambda z: z[1])
#ax.axvline(x[vind], color='green')
#ax.axvline(1.0-0.73, color='green')
lines = l1 + l2 + l3
ax2.set_yscale('log')
ax.set_xlabel('distance threshold')
ax.set_ylabel('Homog. & Compl.')
ax2.set_ylabel('# clusters', rotation=270)
ax.set_ylim([0, 1])
fig.savefig(
D.fmt_path('plotsdir/homogeneity_and_completeness_presentation/' +
category + '2.png'))
def plot_domain_alignment():
'''
all_sets = list()
pool = Pool(6)
for tmp_counts in pool.imap_unordered(get_domain_alignment, range(len(g_clusters))):
if tmp_counts:
all_sets.append(tmp_counts)
try:
with open(g_ca.fmt_path('datadir/domain_alignment/raw.json'),'w') as f:
json.dump(all_sets,f)
except:
print('failed to save raw')
data = list()
means = dict()
for i, cluster in enumerate(all_sets):
alns, sizes, perfs = zip(*cluster.values())
mean_aln = np.mean(alns)
perfs = [z for z in perfs if z > 0]
if perfs:
mean_perf = np.mean(perfs)
else:
mean_perf = None
means[i] = (mean_aln, mean_perf)
for dom, val in cluster.items():
aln, s, perf = val
data.append((dom, aln - mean_aln,
perf - mean_perf if mean_perf and perf else None))
with open(g_ca.fmt_path('datadir/domain_alignment/deviations.json'),'w') as f:
json.dump(data,f)
'''
D = DataGetter()
#with open(D.fmt_path('datadir/domain_alignment/deviations.json'),'r') as f:
with open(D.fmt_path('datadir/deviations.json'), 'r') as f:
data = json.load(f)
doms, aln_devs, perf_devs = zip(*data)
fig, ax = plt.subplots(figsize=(6, 3.5))
ecdf = ECDF(aln_devs)
ax.plot(list(ecdf.x), list(ecdf.y))
ax.set_xlabel('distance from mean alignment')
ax.set_ylabel('CDF')
fig.savefig(D.fmt_path('plotsdir/domain_alignment/alignment.png'))
plt.close(fig)
fig, ax = plt.subplots(figsize=(4.5, 4.5))
aln_devs, perf_devs = zip(
*[z for z in zip(aln_devs, perf_devs) if z[1] is not None])
heatmap, x, y = np.histogram2d(aln_devs, perf_devs, bins=50)
extent = [x[0], x[-1], y[0], y[-1]]
pos = ax.imshow(heatmap.T,
extent=extent,
origin='lower',
cmap='Greys',
aspect='auto')
fig.colorbar(pos)
ax.set_xlabel('distance from mean alignment')
ax.set_ylabel('distance from mean performance')
fig.savefig(D.fmt_path('plotsdir/domain_alignment/align_vs_perf.png'))
plt.close(fig)
def plot_geo_centers():
D = DataGetter()
world = gp.read_file(gp.datasets.get_path('naturalearth_lowres'))
world = world[(world.pop_est > 0) & (world.name != "Antarctica")]
with open(D.fmt_path('datadir/geo_vs_perf/raw.json'), 'r') as f:
geometry, sizes = zip(
*[(Point(z['geo_center_loc'][1], z['geo_center_loc'][0]),
len(z['raw_data'])) for z in json.load(f)])
fig, ax = plt.subplots(figsize=(15, 15))
world.plot(edgecolor='gray', ax=ax)
gdf = gp.GeoDataFrame(geometry=list(geometry))
gdf.plot(ax=ax, markersize=5, color='#00FF00')
fig.savefig(D.fmt_path('plotsdir/geo_centers.png'))
plt.close(fig)
def plot_nearest_centers():
D = DataGetter()
with open(D.fmt_path('datadir/nearest_centers.json'), 'r') as f:
data = json.load(f)
cnres = list()
dists = list()
for item in data:
_, _, c0, c1, d0, d1 = item
cnres.append((1.0 - c0, 1.0 - c1))
dists.append((d0, d1))
x, y = zip(*cnres)
#rng = [minlim, maxlim]
#heatmap, x, y = np.histogram2d(x,y,bins=100, range=[rng,rng])
#extent = [x[0], x[-1], y[0], y[-1]]
fig, ax = plt.subplots(figsize=(4, 6))
ax.scatter(x, y, alpha=0.1)
#minlim = min([ax.get_xlim()[0], ax.get_ylim()[0]])
#maxlim = max([ax.get_xlim()[1], ax.get_ylim()[1]])
xlim = ax.get_xlim()
ylim = ax.get_ylim()
#pos = ax.imshow(heatmap.T, extent=extent, origin='lower', cmap='Greys')
#fig.colorbar(pos)
ax.set_xlabel('CNRE with default center')
ax.set_ylabel('CNRE with closest center')
ax.plot([0, 1], [0, 1], 'r')
ax.set_xlim(xlim)
ax.set_ylim(ylim)
fig.savefig(D.fmt_path('plotsdir/nearest_centers_cnre.png'))
plt.close(fig)
x, y = zip(*dists)
fig, ax = plt.subplots(figsize=(4.5, 4.5))
ax.scatter(x, y, alpha=0.1)
ax.set_xlabel('km to default center')
ax.set_ylabel('km to closest center')
ax.set_xscale('log')
ax.set_yscale('log')
minlim = min([ax.get_xlim()[0], ax.get_ylim()[0]])
maxlim = max([ax.get_xlim()[1], ax.get_ylim()[1]])
ax.set_xlim([minlim, maxlim])
ax.set_ylim([minlim, maxlim])
ax.plot([minlim, maxlim], [minlim, maxlim], 'r')
fig.savefig(D.fmt_path('plotsdir/nearest_centers_dist.png'))
plt.close(fig)
def get_addr_differences(self, cluster):
'''
set of differing answers from a cluster and determine rarity outside of cluster
TODO: make 'flat_answer_counts.pkl', which groups by addr instead of by (site,addr)
'''
answers = defaultdict(set)
for i in cluster:
for site, addrs in self.scb.nodes[i].results:
for addr in addrs:
answers[(site, addr)].add(i)
D = DataGetter()
with open(D.fmt_path('datadir/pkls/flat_answer_counts.pkl'), 'r') as f:
global_counts = pkl.load(f)
uniques = defaultdict(list)
for site, addr in answers:
world = float(global_counts[(site, addr)])
coverage = float(len(answers[(site, addr)])) / world
if coverage >= 0.90:
uniques[site].append(addr)
return uniques
def plot_domain_error(self, *allstats):
D = DataGetter()
with open(D.fmt_path('datadir/domain_error/summary.json'), 'r') as f:
stats = json.load(f)
mean_errs = list()
answers = list()
stds = list()
tups = list()
for dom in stats:
mean_errs.append(stats[dom]['abs_mean'])
stds.append(stats[dom]['std'])
answers.append(stats[dom]['diversity'])
stds = [50.0 * float(z) / float(max(stds)) for z in stds]
fig, (ax) = plt.subplots(1, 1, figsize=(6, 6))
ax.scatter(mean_errs, answers, stds, alpha=0.3)
ax.set_xlabel('mean domain error')
ax.set_ylabel('# distinct answers from domain')
ax.set_yscale('log')
fig.savefig(
self.fmt_path('plotsdir/domain_error/' + self.timeid + '.png'))
plt.close(fig)
print(self.fmt_path('plotsdir/domain_error/' + self.timeid + '.png'))
def condense_domain_error(self):
D = DataGetter(prefix=self.prefix)
data = dict()
for dom in D.test_counts.keys():
uncondensed = list()
with open(D.datadir + 'domain_error/' + str(dom) + '.json',
'r') as f:
for line in f:
uncondensed += json.loads(line)
if len(uncondensed):
condensed = {
'std': np.std(uncondensed),
'raw_mean': np.mean(uncondensed),
'raw_median': np.median(uncondensed),
'diversity': D.diversity(dom)
}
uncondensed = [abs(z) for z in uncondensed]
condensed['abs_mean'] = np.mean(uncondensed)
condensed['abs_median'] = np.median(uncondensed)
condensed['25'] = np.percentile(uncondensed, 25)
condensed['75'] = np.percentile(uncondensed, 75)
data[dom] = condensed
with open(D.fmt_path('datadir/domain_error/summary.json'), 'w') as f:
json.dump(data, f)
def plot_cnre_country_map():
D = DataGetter()
world = gp.read_file(gp.datasets.get_path('naturalearth_lowres'))
world = world[(world.pop_est > 0) & (world.name != "Antarctica")]
names = world.name.tolist()
pos = {names[i]: i for i in range(len(names))}
pos['United States'] = 4
pos['Iran, Islamic Republic of'] = 107
weights = defaultdict(set)
conversions = dict()
with open(D.fmt_path('datadir/countries_codes.csv'), 'r') as f:
for line in f:
pieces = [z.replace('"', '') for z in line.split('",')]
try:
conversions[pieces[1].strip()] = pieces[0].strip()
except:
pass
data = list()
with open(D.fmt_path('datadir/closeness_vs_category/data.json'), 'r') as f:
for line in f:
d = json.loads(line)
data += [z for z in d if z['c'] == 'country']
for item in data:
weights[item['l']] = item['df']
with open(D.fmt_path('datadir/country_cnre_distance_map.json'), 'w') as f:
json.dump(weights, f)
vals = list()
covered = {
pos[conversions[z]]: weights[z]
for z in weights if z and z in conversions and conversions[z] in pos
}
nones = list()
for i in range(len(names)):
if i in covered:
vals.append(covered[i])
else:
vals.append(min(covered.values()))
nones.append(i)
print(sorted(vals))
world['cnre'] = vals
fig, ax = plt.subplots(figsize=(6, 4))
world.plot(column='cnre',
cmap='cool',
edgecolor='gray',
ax=ax,
norm=colors.Normalize(vmin=min(vals), vmax=max(vals)))
for name in nones:
if name > 158:
name += 1
plotCountryPatch(ax, name, 'white', world)
ax.set_xticks([], [])
ax.set_yticks([], [])
sm = plt.cm.ScalarMappable(cmap='cool',
norm=colors.Normalize(vmin=min(vals),
vmax=max(vals)))
sm._A = []
cax = fig.add_axes([0.08, 0.15, 0.82, 0.05])
cbar = fig.colorbar(sm, cax=cax, orientation='horizontal')
cax.set_xlabel('mean external CNRE')
fig.savefig(D.fmt_path('plotsdir/cnre_country_uniqueness_map.png'))
plt.close(fig)
def dump_homogeneity_and_completeness(data):
D = DataGetter()
with open(
D.fmt_path('datadir/homogeneity_and_completeness/' +
data['category'] + '.json'), 'a') as f:
f.write(json.dumps(data) + '\n')
def dump_geo_vs_cnre(data):
D = DataGetter()
with open(D.fmt_path('datadir/vs_cnre/geodists'), 'a') as f:
for res in data:
f.write(json.dumps(res) + '\n')