def results(output_dir="results", analysed_results_dir="obj"):
output_dir = os.path.join(output_dir, analysed_results_dir)
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
os.makedirs(output_dir)
os.makedirs(os.path.join(output_dir, "figures"))
sizes = load_obj("sizes", analysed_results_dir)
nrParticlesInHalos = load_obj("nrParticlesInHalos", analysed_results_dir)
nrHalos = load_obj("nrHalosnrHalos", analysed_results_dir)
percentageInHalos = load_obj("percentageInHalos", analysed_results_dir)
nrParticles = load_obj("nrParticles", analysed_results_dir)
for key in sizes:
f = open(os.path.join(output_dir, key + ".txt"), "w")
f.write(" Mean stderror\n")
f.write("nrParticles: {} {}\n".format(
np.mean(nrParticles[key]), scipy.stats.sem(nrParticles[key])))
f.write("nrHalos: {} {}\n".format(
np.mean(nrHalos[key]), scipy.stats.sem(nrHalos[key])))
f.write("nrParticlesInHalos: {} {}\n".format(
np.mean(nrParticlesInHalos[key]),
scipy.stats.sem(nrParticlesInHalos[key])))
f.write("percentageInHalos: {} {}\n".format(
np.mean(percentageInHalos[key]),
scipy.stats.sem(percentageInHalos[key])))
f.close()
# Plotting all data
for key in sizes:
c = 0
i = 0
t_max = 0
t_min = 10000
data_max = 0
data_min = 10000
legend = []
nr_hues = len(sizes[key])
for data in sizes[key]:
t = range(1, len(data) + 1)
prettyPlot(data,
t,
"Cumulative cluster size",
"Cluster size",
"Nr of clusters",
color=c,
nr_hues=nr_hues,
new_figure=False)
if max(data) > data_max:
data_max = max(data)
if min(data) < data_min:
data_min = min(data)
if max(t) > t_max:
t_max = max(t)
if min(t) < t_min:
t_min = min(t)
legend.append("Datasett {}".format(i))
i += 1
c += 1
plt.yscale('log')
plt.xscale('log')
plt.ylim([t_min, t_max])
plt.xlim([data_min, data_max])
plt.legend(legend)
plt.savefig(os.path.join(output_dir, "figures", key + ".png"))
# plt.clf()
plt.close()
#Calculate and plot mean values
for key in sizes:
size, mean, bins, width = calculateMean(sizes[key])
cumsumAll = np.cumsum(mean[:, ::-1], 1)[:, ::-1]
cumsum = np.mean(cumsumAll, 0)
sumstd = scipy.stats.sem(cumsumAll, 0)
# prettyPlot(size, std, "Cumulative cluster size, mean", "nr cells, mean", "nr of cluster with number of cells > nrParticles", color=2)
# prettyPlot(size, cumsum, "Cumulative cluster size, mean", "nr cells, mean", "nr of cluster with number of cells > nrParticles", color=0, new_figure=False)
# plt.legend(["Mean", "Standard deviation"])
max_sizes = []
for size in sizes[key]:
max_sizes.append(max(size))
max_size = max(max_sizes)
bins = (bins / max_size) * 100
width = (width / max_size) * 100
ax = prettyBar(cumsum,
index=bins[:-1],
color=0,
nr_hues=2,
error=sumstd,
width=width,
linewidth=2)
ax.set_xticks(bins - width / 2)
ax.set_xticklabels(np.round(bins, 0).astype(int),
fontsize=14,
rotation=0)
plt.yscale('log')
plt.ylabel("Nr of clusters")
plt.xlabel("Cluster size [\% of max cluster size {}]".format(max_size),
fontsize=16)
plt.title("Cumulative cluster size, mean")
plt.savefig(os.path.join(output_dir, "figures", key + "mean.png"))
plt.close()
# Calculate fractionalDifference between H and V
pattern = re.compile(r"(.*)(H)$")
for keyH in sizes:
result = pattern.search(keyH)
if result is not None:
keyV = re.sub(pattern, r"\1V", keyH)
if keyV in sizes:
data_max = 0
data_min = 100000
for data in sizes[keyH]:
if max(data) > data_max:
data_max = max(data)
if min(data) < data_min:
data_min = min(data)
for data in sizes[keyV]:
if max(data) > data_max:
data_max = max(data)
if min(data) < data_min:
data_min = min(data)
bins = np.linspace(data_min, data_max, nr_bins)
# plot cumulative cluster size H and V from same animal in the same plot
t_max = 0
t_min = 10000
nr_hues = 2
prettyPlot([data_min - 1], [t_min - 1],
"Cumulative cluster size",
"Cluster size",
"Nr of clusters",
color=0,
nr_hues=nr_hues,
new_figure=False)
prettyPlot([data_min - 1], [t_min - 1],
"Cumulative cluster size",
"Cluster size",
"Nr of clusters",
color=1,
nr_hues=nr_hues,
new_figure=False)
plt.legend(["V", "H"])
for data in sizes[keyV]:
t = range(1, len(data) + 1)
prettyPlot(data,
t,
"Cumulative cluster size",
"Cluster size",
"Nr of clusters",
color=0,
nr_hues=nr_hues,
new_figure=False)
if max(t) > t_max:
t_max = max(t)
if min(t) < t_min:
t_min = min(t)
i += 1
for data in sizes[keyH]:
t = range(1, len(data) + 1)
prettyPlot(data,
t,
"Cumulative cluster size",
"Cluster size",
"Nr of clusters",
color=1,
nr_hues=nr_hues,
new_figure=False)
if max(t) > t_max:
t_max = max(t)
if min(t) < t_min:
t_min = min(t)
i += 1
plt.yscale('log')
plt.xscale('log')
plt.ylim([t_min, t_max])
plt.xlim([data_min, data_max])
plt.savefig(
os.path.join(output_dir, "figures",
keyH[:-1] + "combined.png"))
plt.close()
meanH = []
for data in sizes[keyH]:
meanH.append(np.histogram(data, bins=bins)[0])
meanH = np.array(meanH)
meanV = []
for data in sizes[keyV]:
meanV.append(np.histogram(data, bins=bins)[0])
meanV = np.array(meanV)
width = bins[1] - bins[0]
cumsumAllH = np.cumsum(meanH[:, ::-1], 1)[:, ::-1]
cumsumAllV = np.cumsum(meanV[:, ::-1], 1)[:, ::-1]
cumsumH = np.mean(cumsumAllH, 0)
cumsumV = np.mean(cumsumAllV, 0)
sumstdV = np.std(cumsumAllV, 0)
sumstdH = np.std(cumsumAllH, 0)
diff = 1 - cumsumH / cumsumV.astype(float)
stddiff = diff * np.sqrt(sumstdV**2 / cumsumV**2 +
sumstdH**2 / cumsumH**2)
#plot two mean values against each other
bins = (bins / data_max) * 100
width = bins[1] - bins[0]
ax = prettyBar(cumsumV,
index=bins[:-1],
color=0,
nr_hues=2,
error=sumstdV,
width=width,
linewidth=2)
ax = prettyBar(cumsumH,
index=bins[:-1],
color=4,
nr_hues=6,
error=sumstdH,
width=width,
linewidth=2,
new_figure=False,
alpha=0.6,
error_kw=dict(ecolor=get_colormap()[4],
lw=2,
capsize=10,
capthick=2))
ax.set_xticks(bins - width / 2)
ax.set_xticklabels(np.round(bins, 0).astype(int),
fontsize=14,
rotation=0)
plt.yscale('log')
plt.legend(["V", "H"])
plt.ylabel("Nr of clusters")
plt.xlabel("Cluster size [\% of max cluster size {}]".format(
data_max),
fontsize=16)
plt.title("Cumulative cluster size, mean")
plt.savefig(
os.path.join(output_dir, "figures",
keyH[:-1] + "compare.png"))
plt.close()
# Plot fractional difference
width = bins[1] - bins[0]
ax = prettyBar(diff,
index=bins[:-1],
color=0,
nr_hues=2,
error=stddiff,
width=width,
linewidth=2)
ax.set_xticks(bins - width / 2)
ax.set_xticklabels(np.round(bins, 0).astype(int),
fontsize=14,
rotation=0)
plt.ylabel("Fractional difference nr of cluster")
plt.xlabel("Cluster size [\% of max cluster size {}]".format(
data_max),
fontsize=16)
plt.title("Fractional difference, (V-H)/V")
# prettyPlot(size, diff, "Fractional difference, (V-H)/V", "CLuster size", "Fractional difference nr of cluster", color=0)
plt.savefig(
os.path.join(output_dir, "figures",
keyH[:-1] + "difference.png"))
plt.close()
width = 0.2
distance = 0.5
xlabels = ["Mean", "Variance", "$P_5$", "$P_{95}$"]
xticks = [0, width, distance + width, distance + 2 * width]
values = [
data["nr_spikes"].mean, data["nr_spikes"].variance,
data["nr_spikes"].percentile_5, data["nr_spikes"].percentile_95
]
ylabel = data.get_labels("nr_spikes")[0]
ax = prettyBar(values,
index=xticks,
xlabels=xlabels,
ylabel=ylabel.capitalize(),
palette=get_colormap_tableu20(),
style="seaborn-white")
plt.savefig("nr_spikes.png")
xlabels = [
r"Potassium conductance $\bar{g}_\mathrm{K}$",
r"Sodium conductance $\bar{g}_\mathrm{Na}$",
r"Leak conductance $\bar{g}_\mathrm{l}$"
]
xticks = [0, width + 0.1, 2 * (width + 0.1)]
values = [
data["nr_spikes"].mean, data["nr_spikes"].variance,
def test_prettyPlotFalseNewFigure2(self):
prettyBar(self.values, new_figure=False)
plt.close()
def test_PrettyBarLabels(self):
prettyBar(self.values, xlabels=self.labels)
plt.close()
def test_PrettyBarError(self):
prettyBar(self.values, self.error)
plt.close()
def test_PrettyBar(self):
prettyBar(self.values)
plt.close()