lw = 2
wlen = 7
wtype = "hanning"
plt.figure(figsize=(12, 4))
plt.plot(xdata, [54]*len(xdata), "-", label="Chance level", color="#555753", linewidth=lw)
plt.rcParams['font.size'] = 10
plt.rcParams['font.family'] = "Ubuntu"
plt.suptitle("Results")
for col in [2, 5, 8, 11]:
ydata = np.array(d[:,col], dtype=float)
yerr = np.array(d[:,col+1], dtype=float)
_smooth = smooth(ydata, window_len=2*wlen+1, window=wtype)[wlen:-wlen]
color = colors.pop()
color2 = colors2.pop()
plt.errorbar(xdata, ydata*100, yerr=yerr*100, fmt=",", color=color2, capsize=0, elinewidth=0.5)
plt.plot(xdata, _smooth*100, "-", label=labels.pop(), color=color, linewidth=lw)
plt.ylim(50, 104)
plt.legend(fancybox=True, borderpad=1)
plt.xlabel('Noise (sigma)')
plt.ylabel('Clustering accuracy (%)')
plt.grid(True, color=(0,0,0))
plt.savefig("plot.svg")
plt.show()
import matplotlib.pyplot as plt
from matplotlib import cm
from plotting import smooth, heatcurve
wlen = 7
d = np.loadtxt("summary.csv", dtype=str, skiprows=1, delimiter=", ")
for col in [38, 41, 47]:
xdata = np.array(d[:,0], dtype=float)
ydata = np.array(d[:,col], dtype=float)*100
yerr = np.array(d[:,col+1], dtype=float)*100
a = np.swapaxes([ydata, yerr], 0, 1)
_smooth = smooth(ydata, window_len=2*wlen+1)[wlen:-wlen]
bfl = slider(a, wlen=wlen, lower=0, upper=46)
fig = heatcurve(xdata, ydata, cdata=_smooth, yerr=yerr, yheat=np.log10(bfl),
heatrange=(-3,3), threshold=2, size=(12,4), heatlabel="Log10(Bf)",
ticks=range(-3,4))
plt.xlabel('Noise (sigma)', figure=fig)
plt.ylabel('Difference in clustering accuracy (%)', figure=fig)
if col == 47:
plt.ylim(-5, 50, figure=fig)
else:
plt.ylim(-5, 25, figure=fig)
plt.rcParams['font.size'] = 10
plt.rcParams['font.family'] = "Ubuntu"
indy = k
label = "df:" + str(df)
if wrt == "lr":
indx = i
indy = k
label = "lr:" + str(lr)
if wrt == "ms":
indx = i
indy = j
label = "ms:" + str(ms)
try:
filename = file_names[file_names_short.index(namee)]
dataframe = pd.read_csv(os.path.join(path, filename))
values = dataframe["Value"].tolist()
smoothed_values = smooth(values, smoothing_factor)
times = np.array(dataframe["Wall time"])[:len(smoothed_values)]/6000.
times -= times[0] # Make it relative
name = filename.replace("run-CartPole-v0_", "").split("_2019")[0]
name = name.replace("_", " ").replace("-", ":")
# Plot in corresponding axes
ax[indy][indx].plot(times, smoothed_values, label=label)
# ax[indx][indy].set_title(title)
except Exception as e:
print(e)
# print("Mising:")
# print(filename)
# plt.set_xlabel('Relative Time [hours]')
# plt.set_ylabel('Score [Points]')