def __init__(self, score, related_data={}):
if not isinstance(score, Exception) and not isinstance(score, float):
raise sciunit.InvalidScoreError("Score must be a float.")
elif score < 0.0 or score > 100.0:
raise sciunit.InvalidScoreError(("Score of %f must be in "
"range 0.0-100.0" % score))
else:
super(PercentScore, self).__init__(score,
related_data=related_data)
def __init__(self, score, related_data={}):
if not isinstance(score, Exception) and \
not isinstance(score, float) and \
not (isinstance(score, pq.Quantity) and score.size==1):
raise sciunit.InvalidScoreError("Score must be a float.")
else:
super(FloatScore, self).__init__(score, related_data=related_data)
def compute_score(self, observation, prediction, verbose=False):
"""Implementation of sciunit.Test.score_prediction."""
try:
assert len(observation) == len(prediction)
except Exception:
# or return InsufficientDataScore ??
raise sciunit.InvalidScoreError(
("Difference in # of layers."
" Cannot continue test for layer heights."))
zscores = {}
for key0 in observation.keys():
zscores[key0] = sciunit.scores.ZScore.compute(
observation[key0]["height"], prediction[key0]["height"]).score
self.score = morphounit.scores.CombineZScores.compute(zscores.values())
self.score_dict = zscores
self.score.description = "Mean of absolute Z-scores"
# create output directory
self.path_test_output = self.directory_output + 'layer_height/' + self.model_name + '/'
if not os.path.exists(self.path_test_output):
os.makedirs(self.path_test_output)
self.observation = observation
self.prediction = prediction
# create relevant output files
# 1. Error Plot
err_plot = plots.ErrorPlot(self)
err_plot.xlabels = OrderedDict(
sorted(self.observation.items(), key=lambda t: t[0]))
err_plot.ylabel = "Layer Height (um)"
file1 = err_plot.create()
self.figures.append(file1)
# 2. Text Table
txt_table = plots.TxtTable(self)
file2 = txt_table.create(mid_keys=["height"])
self.figures.append(file2)
"""
# save document with Z-score data
filename = path_test_output + 'score_summary' + '.txt'
dataFile = open(filename, 'w')
dataFile.write("==============================================================================\n")
dataFile.write("Test Name: %s\n" % self.name)
dataFile.write("Model Name: %s\n" % self.model_name)
dataFile.write("------------------------------------------------------------------------------\n")
dataFile.write("Layer #\tExpt. mean\tExpt. std\tModel value\tZ-score\n")
dataFile.write("..............................................................................\n")
for key0 in zscores.keys():
o_mean = observation[key0]["height"]["mean"]
o_std = observation[key0]["height"]["std"]
p_value = prediction[key0]["height"]["value"]
dataFile.write("%s\t%s\t%s\t%s\t%s\n" % (key0, o_mean, o_std, p_value, zscores[key0]))
dataFile.write("------------------------------------------------------------------------------\n")
dataFile.write("Combined Score: %s\n" % score)
dataFile.write("==============================================================================\n")
dataFile.close()
self.figures.append(filename)
"""
return self.score
def compute_score(self, observation, prediction, verbose=True):
"""Implementation of sciunit.Test.score_prediction"""
assert len(observation) == len(prediction), \
sciunit.InvalidScoreError(("Difference in # of m-type cells. Cannot continue test"
"for laminar distribution of synapses across CA1 layers"))
# print "observation = ", observation, "\n"
# print "prediction = ", prediction, "\n"
# Computing the score
scores_cell = dict.fromkeys(observation.keys(),[])
for key0 in scores_cell.keys(): # m-type cell (AA, BP, BS, CCKBC, Ivy, OLM, PC, PPA, SCA, Tri)
scores_cell[key0] = getattr(hpn_scores, score_str).compute(observation[key0], prediction[key0])
# create output directory
path_test_output = self.directory_output + self.model_name + '/'
if not os.path.exists(path_test_output):
os.makedirs(path_test_output)
# save figure with score data
scores_cell_floats = dict.fromkeys(observation.keys(), [])
for key0, score_cell in scores_cell.items():
scores_cell_floats[key0] = [score_cell.score.statistic_n, score_cell.score.pvalue]
score_label = score_str[:-5] + '-score'
scores_cell_df = pd.DataFrame(scores_cell_floats, index=[score_label, 'p-value'])
scores_cell_df = scores_cell_df.transpose()
print scores_cell_df, '\n'
# pal = sns.cubehelix_palette(len(observation))
pal = sns.color_palette('Reds', len(observation))
rank = [int(value)-1 for value in scores_cell_df[score_label].rank()]
axis_obj = sns.barplot(x=scores_cell_df[score_label], y=scores_cell_df.index, palette=np.array(pal)[rank])
axis_obj.set(xlabel=score_label, ylabel='Cell')
sns.despine()
for i, p in enumerate(axis_obj.patches):
axis_obj.annotate("p = %.2f" % scores_cell_df['p-value'].values[i],
xy=(p.get_x() + p.get_width(), p.get_y() + 0.5),
xytext=(3, 0), textcoords='offset points')
filename = path_test_output + score_str + '_plot' + '.png'
plt.savefig(filename, dpi=600,)
self.figures.append(filename)
# self.score = morphounit.scores.CombineZScores.compute(zscores.values())
self.score = scores_cell_floats["PC"][0]
return hpn_scores.FreemanTukey2Score(self.score)
def compute_score(self, observation, prediction, verbose=True):
"""Implementation of sciunit.Test.score_prediction"""
assert len(observation) == len(prediction), \
sciunit.InvalidScoreError(("Difference in # of m-type cells. Cannot continue test"
"for laminar distribution of synapses across CA1 layers"))
# print "observation = ", observation, "\n"
# print "prediction = ", prediction, "\n"
# Computing the score
scores_cell = dict.fromkeys(observation.keys(),[])
for key0 in scores_cell.keys(): # m-type cell (AA, BP, BS, CCKBC, Ivy, OLM, PC, PPA, SCA, Tri)
scores_cell[key0] = getattr(hpn_scores, score_str).compute(observation[key0], prediction[key0])
# create output directory
path_test_output = self.directory_output + self.model_name + '/'
if not os.path.exists(path_test_output):
os.makedirs(path_test_output)
# save figure with score data
scores_cell_floats = dict.fromkeys(observation.keys(), [])
for key0, score_cell in scores_cell.items():
scores_cell_floats[key0] = [score_cell.score]
score_label = score_str[:-5] + '-score'
scores_cell_df = pd.DataFrame(scores_cell_floats, index=[score_label])
scores_cell_df = scores_cell_df.transpose()
print scores_cell_df, '\n'
# pal = sns.cubehelix_palette(len(observation))
pal = sns.color_palette('Reds', len(observation))
rank = [int(value)-1 for value in scores_cell_df[score_label].rank()]
axis_obj = sns.barplot(x=scores_cell_df[score_label], y=scores_cell_df.index, palette=np.array(pal)[rank])
axis_obj.set(xlabel=score_label, ylabel='Cell')
sns.despine()
filename = path_test_output + score_str + '_plot' + '.pdf'
plt.savefig(filename, dpi=600,)
self.figures.append(filename)
scores_array = scores_cell_df[score_label].array
self.score = sum(map(abs,scores_array)) / len(scores_array)
return hpn_scores.KLdivScore(self.score)
def __init__(self, score, related_data={}):
if not isinstance(score, Exception) and not isinstance(score, float):
raise sciunit.InvalidScoreError("Score must be a float.")
else:
super(ZScore, self).__init__(score, related_data=related_data)
def __init__(self, score, related_data={}):
if isinstance(score, Exception) or score in [True, False]:
super(BooleanScore, self).__init__(score,
related_data=related_data)
else:
raise sciunit.InvalidScoreError("Score must be True or False.")
