def hist_from_file(input_file, header, threshold, range_min, range_max, stats,
markers, columns, vert_space):
""" Gets comma separated key-value pairs, and plots hist of the values for every key
(given more than threshold occurrences) """
values = defaultdict(list)
fp = csv.reader(input_file)
h = [None, "Values"
] # Init h so that h[1] will hold the xlabel header or no header
if header:
h = fp.next()
min_value = float(range_min) if range_min else None
max_value = float(range_max) if range_max else None
for l in fp:
if len(l) != 2:
raise ValueError("Bad number of items in line: " + str(l))
(k, v) = (l[0], float(l[1]))
values[k].append(v)
# Update minimum and maximum values, later to be used in the range of the histograms
if not range_min: # only if the range wasn't given as a parameter
if not min_value:
min_value = v
elif min_value > v:
min_value = v
if not range_max:
if not max_value:
max_value = v
elif max_value < v:
max_value = v
plots = sum([len(v) > threshold for v in values.itervalues()])
i = 1
plt.figure()
for k, v in values.iteritems():
if len(v) > threshold:
plt.subplot(ceil(plots / float(columns)), int(columns), i)
if stats:
k += "\nm=%0.2f,s=%0.2f,med=%0.2f" % (mean(v), std(v),
median(v))
draw_hist(k,
v, (min_value, max_value),
ylabel="Count",
xlabel=h[1])
if markers:
plt.axvline(mean(v), color='r', linestyle='-')
plt.axvline(mean(v) + std(v), color='r', linestyle='--')
plt.axvline(mean(v) - std(v), color='r', linestyle='--')
plt.axvline(median(v), color='b', linestyle='-')
print k.replace('\n', ': \t') + ", Count:" + str(len(v))
i += 1
plt.subplots_adjust(hspace=float(
vert_space)) # tweak spacing to prevent clipping of x-label
plt.show()
def hist_from_file(input_file, header, threshold, range_min, range_max, stats, markers, columns, vert_space):
""" Gets comma separated key-value pairs, and plots hist of the values for every key
(given more than threshold occurrences) """
values = defaultdict(list)
fp = csv.reader(input_file)
h = [None, "Values"] # Init h so that h[1] will hold the xlabel header or no header
if header:
h = fp.next()
min_value = float(range_min) if range_min else None
max_value = float(range_max) if range_max else None
for l in fp:
if len(l) != 2:
raise ValueError("Bad number of items in line: " + str(l))
(k, v) = (l[0], float(l[1]))
values[k].append(v)
# Update minimum and maximum values, later to be used in the range of the histograms
if not range_min: # only if the range wasn't given as a parameter
if not min_value:
min_value = v
elif min_value > v:
min_value = v
if not range_max:
if not max_value:
max_value = v
elif max_value < v:
max_value = v
plots = sum([len(v) > threshold for v in values.itervalues()])
i = 1
plt.figure()
for k, v in values.iteritems():
if len(v) > threshold:
plt.subplot(ceil(plots / float(columns)), int(columns), i)
if stats:
k += "\nm=%0.2f,s=%0.2f,med=%0.2f" % (mean(v), std(v), median(v))
draw_hist(k, v, (min_value, max_value), ylabel="Count", xlabel=h[1])
if markers:
plt.axvline(mean(v), color='r', linestyle='-')
plt.axvline(mean(v)+std(v), color='r', linestyle='--')
plt.axvline(mean(v)-std(v), color='r', linestyle='--')
plt.axvline(median(v), color='b', linestyle='-')
print k.replace('\n', ': \t') + ", Count:" + str(len(v))
i += 1
plt.subplots_adjust(hspace=float(vert_space)) # tweak spacing to prevent clipping of x-label
plt.show()
def draw_distributions(input):
accuracies = []
per_firm_accuracy = defaultdict(list)
for firm, accuracy in analyst_accuracies(input):
accuracies.append(accuracy)
per_firm_accuracy[firm].append(accuracy)
p = plt.figure()
p.canvas.set_window_title("All Analysts")
draw_hist("All Analysts",accuracies, range=(0,2.0))
p = plt.figure()
plots = sum([len(v) > 200 for v in per_firm_accuracy.itervalues()])
i = 1
for firm,values in per_firm_accuracy.iteritems():
if len(values) > 200:
plt.subplot(4, ceil(plots / 4.0), i)
draw_hist(firm, values, range=(0,2.0))
i += 1
plt.suptitle('Count vs. Accuracy (|New Price - Target| / |Original Price|)')
p.canvas.set_window_title("Each Analyst Separately")
plt.show()
def draw_distributions(input):
accuracies = []
per_firm_accuracy = defaultdict(list)
for firm, accuracy in analyst_accuracies(input):
accuracies.append(accuracy)
per_firm_accuracy[firm].append(accuracy)
p = plt.figure()
p.canvas.set_window_title("All Analysts")
draw_hist("All Analysts", accuracies, range=(0, 2.0))
p = plt.figure()
plots = sum([len(v) > 200 for v in per_firm_accuracy.itervalues()])
i = 1
for firm, values in per_firm_accuracy.iteritems():
if len(values) > 200:
plt.subplot(4, ceil(plots / 4.0), i)
draw_hist(firm, values, range=(0, 2.0))
i += 1
plt.suptitle(
'Count vs. Accuracy (|New Price - Target| / |Original Price|)')
p.canvas.set_window_title("Each Analyst Separately")
plt.show()