def plot_diadem_score(neuron_distance_csv, outputDir,algorithms=None):
df_nd = pd.read_csv(neuron_distance_csv)
if not path.exists(outputDir):
os.mkdir(outputDir)
if algorithms is None:
algorithms= order_algorithms_by_size(df_nd)
### all algorithm plot
dfg = df_nd.groupby('algorithm')
sample_size_per_algorithm=[]
for alg in algorithms:
sample_size_per_algorithm.append(dfg.get_group(alg).shape[0])
#plot the average node distances
plt.figure()
sb.set_context("talk", font_scale=0.7)
a=sb.barplot(y='algorithm', x='diadem_score', data=df_nd,order=algorithms)
algorithm_name_mapping = rp.get_algorithm_name_dict()
algorithm_names = [algorithm_name_mapping[x] for x in algorithms]
a.set_yticklabels(['%s ($n$=%d )'%(algorithm_names[i], sample_size_per_algorithm[i]) for i in range(algorithms.size) ])
#sb.set_context("talk", font_scale=3.0)
#plt.xticks(rotation="90")
plt.xlabel('Diadem Scores')
plt.subplots_adjust(left=0.5,right=0.95,bottom=0.1, top=0.9)
plt.savefig(outputDir + '/Diadem_score.png', format='png')
#plt.show()
plt.close()
return
def plot_all_score(neuron_distance_csv, outputDir, algorithms=None):
df_nd = pd.read_csv(neuron_distance_csv)
if not path.exists(outputDir):
os.mkdir(outputDir)
if algorithms is None:
algorithms = order_algorithms_by_size(df_nd)
### all algorithm plot
dfg = df_nd.groupby('algorithm')
sample_size_per_algorithm = []
for alg in algorithms:
sample_size_per_algorithm.append(dfg.get_group(alg).shape[0])
#plot the average node distances
plt.figure()
sb.set_style("white")
#g = sb.lmplot(x="image_id", y="diadem_score", hue="algorithm", data=df_nd,fit_reg=False)
a = sb.barplot(y='algorithm',
x=['diadem_score', 'nblast_bi_score'],
data=df_nd,
order=algorithms)
algorithm_name_mapping = rp.get_algorithm_name_dict()
algorithm_names = [algorithm_name_mapping[x] for x in algorithms]
a.set_yticklabels([
'%s ($n$=%d )' % (algorithm_names[i], sample_size_per_algorithm[i])
for i in range(algorithms.size)
])
sb.set_context("talk", font_scale=3.0)
plt.xticks(rotation="90")
plt.xlabel('All Scores')
plt.subplots_adjust(left=0.5, right=0.95, bottom=0.1, top=0.9)
plt.savefig(outputDir + '/All_scores.png', format='png')
plt.show()
# plt.close()
return
def plot_similarities(neuron_distance_csv, outputDir,algorithms=None,metric='neuron_distance',CASE_BY_CASE_PLOT = 0, value_label=None):
df_nd_ori = pd.read_csv(neuron_distance_csv)
algorithm_name_mapping = rp.get_algorithm_name_dict()
df_nd = calculate_similarities(neuron_distance_csv,metric,output_similarity_csv=neuron_distance_csv+".similarity.csv")
all_images = np.unique(df_nd.image_id)
if not path.exists(outputDir):
os.mkdir(outputDir)
if algorithms is None:
algorithms= order_algorithms_by_size(df_nd_ori)
if CASE_BY_CASE_PLOT:
dfg = df_nd.groupby('image_id')
#sample_size_per_algorithm=[]
for image in all_images:
df_image_cur = dfg.get_group(image)
if df_image_cur.shape[0] > 0:
plt.figure()
plt.bar(range(df_image_cur.swc_file.size), df_image_cur['similarity'])
algorithm_names = [algorithm_name_mapping[x] for x in df_image_cur['algorithm']]
plt.xticks(range(df_image_cur.swc_file.size), algorithm_names,
rotation="90")
plt.ylabel(' Similarity (0~1) by ' +metric)
#plt.subplots_adjust(bottom=0.3)
plt.savefig(outputDir + '/sorted/figs/' + image.split('/')[-1] + '_'+metric+'_similarity.png', format='png')
#plt.show()
plt.close()
else:
print image+" has no valid reconstructions"
# dfg = df_nd.groupby('algorithm')
# rate_per_algorithm=[]
# for alg in algorithms:
# df_a = dfg.get_group(alg)
# sucessrate= float(np.count_nonzero(df_a['similarity']))/df_a.shape[0] * 100
# number = np.count_nonzero(df_a['similarity'])
# rate_per_algorithm.append(number)
dfg = df_nd_ori.groupby('algorithm')
sample_size_per_algorithm=[]
for alg in algorithms:
sample_size_per_algorithm.append(dfg.get_group(alg).shape[0])
plt.figure()
sb.set_context("talk", font_scale=0.7)
a=sb.barplot(y='algorithm', x='similarity', data=df_nd,order=algorithms)
algorithm_names = [algorithm_name_mapping[x] for x in algorithms]
a.set_yticklabels(['%s ($n$=%d )'%(algorithm_names[i], sample_size_per_algorithm[i]) for i in range(algorithms.size) ])
#sb.set_context("talk", font_scale=3.0)
#plt.xticks(rotation="90")
if value_label == None:
value_label = ' Similarity (0~1) by '+ metric
plt.ylabel('algorithm (n = # recons)')
plt.xlabel(value_label)
plt.subplots_adjust(left=0.4,right=0.9, bottom=0.1, top=0.9)
plt.savefig(outputDir + '/'+value_label+'.png', format='png')
#plt.show()
plt.close()
return
def plot_similarities(neuron_distance_csv,
outputDir,
algorithms=None,
metric='neuron_distance',
CASE_BY_CASE_PLOT=0,
value_label=None):
df_nd_ori = pd.read_csv(neuron_distance_csv)
algorithm_name_mapping = rp.get_algorithm_name_dict()
df_nd = calculate_similarities(neuron_distance_csv,
metric,
output_similarity_csv=neuron_distance_csv +
".similarity.csv")
all_images = np.unique(df_nd.image_id)
if not path.exists(outputDir):
os.mkdir(outputDir)
if algorithms is None:
algorithms = order_algorithms_by_size(df_nd_ori)
if CASE_BY_CASE_PLOT:
dfg = df_nd.groupby('image_id')
#sample_size_per_algorithm=[]
for image in all_images:
df_image_cur = dfg.get_group(image)
if df_image_cur.shape[0] > 0:
plt.figure()
plt.bar(range(df_image_cur.swc_file.size),
df_image_cur['similarity'])
algorithm_names = [
algorithm_name_mapping[x]
for x in df_image_cur['algorithm']
]
plt.xticks(range(df_image_cur.swc_file.size),
algorithm_names,
rotation="90")
plt.ylabel(' Similarity (0~1) by ' + metric)
#plt.subplots_adjust(bottom=0.3)
plt.savefig(outputDir + '/sorted/figs/' +
image.split('/')[-1] + '_' + metric +
'_similarity.png',
format='png')
#plt.show()
plt.close()
else:
print image + " has no valid reconstructions"
# dfg = df_nd.groupby('algorithm')
# rate_per_algorithm=[]
# for alg in algorithms:
# df_a = dfg.get_group(alg)
# sucessrate= float(np.count_nonzero(df_a['similarity']))/df_a.shape[0] * 100
# number = np.count_nonzero(df_a['similarity'])
# rate_per_algorithm.append(number)
dfg = df_nd_ori.groupby('algorithm')
sample_size_per_algorithm = []
for alg in algorithms:
sample_size_per_algorithm.append(dfg.get_group(alg).shape[0])
plt.figure()
sb.set_context("talk", font_scale=0.7)
a = sb.barplot(y='algorithm', x='similarity', data=df_nd, order=algorithms)
algorithm_names = [algorithm_name_mapping[x] for x in algorithms]
a.set_yticklabels([
'%s ($n$=%d )' % (algorithm_names[i], sample_size_per_algorithm[i])
for i in range(algorithms.size)
])
#sb.set_context("talk", font_scale=3.0)
#plt.xticks(rotation="90")
if value_label == None:
value_label = ' Similarity (0~1) by ' + metric
plt.ylabel('algorithm (n = # recons)')
plt.xlabel(value_label)
plt.subplots_adjust(left=0.4, right=0.9, bottom=0.1, top=0.9)
plt.savefig(outputDir + '/' + value_label + '.png', format='png')
#plt.show()
plt.close()
return
