def plot(ax, data, control_condition, color, dot_size=50, show_ylabel=True):
"""
Produce the control sigmoid figures
"""
x_label = "$EV_{right} - EV_{left}$"
y_label = "p(choose right)"
cd = control_condition
d = data[cd]
scatter_and_sigmoid(x=d['x'],
y=d['y'],
x_fit=d['fit']['x'],
y_fit=d['fit']['y'],
ax=ax,
color=color,
dot_size=dot_size)
# title = LABELS_CONTROL[cd]
# ax.set_title(title, fontsize=axis_label_font_size*1.2)
ax.axhline(0.5,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.axvline(0.0,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.set_ylim(0.0, 1.0)
ax.set_yticks((0, 0.5, 1))
# Axis labels
ax.set_xlabel(x_label)
if show_ylabel:
ax.set_ylabel(y_label)
# Remove top and right borders
ax.spines['right'].set_color('none')
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.spines['top'].set_color('none')
txt = "$k=" + f"{data[cd]['fit'][SIG_STEEP]:.2f}" + "$" + "\n" \
+ "$x_0=" + f"{data[cd]['fit'][SIG_MID]:.2f}" + "$"
# txt = \
# r"$F(x) = \dfrac{1}{1 + \exp(-k (x - x_0))}$" + "\n\n" \
# + "$k=" + f"{data[cd]['fit'][SIG_STEEP]:.2f}" + "$" + "\n" \
# + "$x_0=" + f"{data[cd]['fit'][SIG_MID]:.2f}" + "$" + "\n"
add_text(ax, txt)
def plot(ax, data, color='C0', show_ylabel=True, dot_size=50):
"""
Produce the figure which presents at which extent
the risky option was chosen according to the difference
between expected values,
i.e. the certainty-risk trade-off figure
"""
scatter_and_sigmoid(x=data['x'],
y=data['y'],
x_fit=data['fit']['x'],
y_fit=data['fit']['y'],
color=color,
ax=ax,
dot_size=dot_size)
ax.axhline(0.5,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.axvline(0,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.set_ylim(0.0, 1.0)
ax.set_yticks((0, 0.5, 1))
# Axis labels
ax.set_xlabel(r"$EV_{riskiest} - EV_{safest}$")
if show_ylabel:
ax.set_ylabel("p(choose riskiest)")
# Remove top and right borders
ax.spines['right'].set_color('none')
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.spines['top'].set_color('none')
txt = "$k=" + f"{data['fit'][SIG_STEEP]:.2f}" + "$" + "\n"\
+ "$x_0=" + f"{data['fit'][SIG_MID]:.2f}" + "$" \
# txt = \
# r"$F(x) = \dfrac{1}{1 + \exp(-k (x - x_0))}$" + "\n\n" \
# + "$k=" + f"{data['fit'][SIG_STEEP]:.2f}" + "$" + "\n"\
# + "$x_0=" + f"{data['fit'][SIG_MID]:.2f}" + "$" + "\n" \
add_text(ax, txt)
def plot(ax, data, linestyles=None, color='C0'):
"""
Produce the probability distortion figure
"""
fit_distortion = data["distortion"]
class_model = data["class_model"]
if linestyles is None:
linestyles = ['-' for _ in range(len(fit_distortion))]
for j in range(len(fit_distortion)):
_line(class_model=class_model,
param=fit_distortion[j],
color=color,
alpha=0.5,
linewidth=1,
linestyle=linestyles[j],
ax=ax)
v_mean = np.mean(fit_distortion)
v_std = np.std(fit_distortion)
_line(class_model=class_model,
param=v_mean,
linewidth=3,
color=color,
ax=ax)
add_text(ax, r'$\alpha=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$')
ax.set_xlabel('$p$')
ax.set_ylabel('$w(p)$')
ax.set_ylim(0, 1)
ax.set_xlim(-0.01, 1.01)
ax.set_aspect('equal')
ax.spines['right'].set_color('none')
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.spines['top'].set_color('none')
ax.plot((0, 1), (0, 1),
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.set_xticks((0, 0.5, 1))
ax.set_yticks((0, 0.5, 1))
ax.tick_params(axis='both', which='major')
def plot(ax, data, linestyles=None, color='C0', alpha_chunk=0.5):
"""
Produce the utility function figure
"""
pr = data['risk_aversion']
class_model = data['class_model']
cond = data['cond']
if linestyles is None:
linestyles = ['-' for _ in range(len(pr))]
if cond == GAIN:
x = np.linspace(0, 1, 1000)
ax.set_xlim(0, 1)
ax.set_ylim(0, 1)
ax.set_xticks([0, 1])
ax.set_yticks([0, 1])
ax.plot((0, 1), (0, 1),
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
elif cond == LOSS:
x = np.linspace(-1, 0, 1000)
ax.set_xlim(-1, 0)
ax.set_ylim(-1, 0)
ax.set_xticks([-1, 0])
ax.set_yticks([-1, 0])
ax.plot((-1, 0), (-1, 0),
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
else:
raise ValueError
for j in range(len(pr)):
_line(x=x,
class_model=class_model,
risk_aversion=pr[j],
color=color,
ax=ax,
linewidth=1,
alpha=alpha_chunk,
linestyle=linestyles[j])
v_mean = np.mean(pr)
v_std = np.std(pr)
_line(x=x,
risk_aversion=v_mean,
class_model=class_model,
ax=ax,
color=color)
add_text(ax, r'$\beta=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$')
ax.spines['right'].set_color('none')
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.spines['top'].set_color('none')
ax.set_xlabel("$x$")
ax.set_ylabel("$u(x)$")
ax.set_aspect(1)
def plot(ax, data, linestyles=None, color='C0'):
"""
Produce the precision figure
"""
n_points = 1000
v_mean = np.mean(data['precision'])
v_std = np.std(data['precision'])
n_chunk = len(data['precision'])
class_model = data['class_model']
if class_model.__name__ == "DMSciReports":
p0 = 0.5
p1, x1 = 1., 0.25
x0_equal_ev = x1 * (1 / p0)
x0_list = np.linspace(x1 + 0.01, 1.00, n_points)
x = x0_list / x1
pairs = []
for i, x0 in enumerate(x0_list):
pairs.append({"p0": p0, "x0": x0, "p1": p1, "x1": x1})
y = np.zeros((n_chunk, len(x)))
for i_c in range(n_chunk):
dm = class_model([data[k][i_c] for k in class_model.param_labels])
for i_p, p in enumerate(pairs):
y[i_c, i_p] = dm.p_choice(c=0, **p)
dm = class_model([np.mean(data[k]) for k in class_model.param_labels])
y_mean = np.zeros(len(x))
for i_p, p in enumerate(pairs):
y_mean[i_p] = dm.p_choice(c=0, **p)
ax.axvline(x0_equal_ev / x1,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
x_label = r"$\frac{x_{risky}}{x_{safe}}$"
y_label = "P(Choose risky option)"
text = r'$\lambda=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$'
elif class_model.__name__ == "AgentSoftmax":
fit_precision = data['precision']
x = np.linspace(-1, 1, n_points)
y = np.zeros((n_chunk, len(x)))
for i_c in range(n_chunk):
v = fit_precision[i_c]
y[i_c] = class_model.softmax(x, v)
y_mean = np.zeros(len(x))
y_mean[:] = class_model.softmax(x, v_mean)
ax.axvline(0,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
x_label = r"$SEU(L_{1}) - SEU(L_{2})$"
y_label = "$P(Choose L_{1})$"
text = r'$\lambda=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$'
elif class_model.__name__ == "AgentSideAdditive":
fit_precision = data['precision']
fit_side_bias = data['side_bias']
x = np.linspace(-1, 1, n_points)
y = np.zeros((n_chunk, len(x)))
for i_c in range(n_chunk):
v = fit_precision[i_c]
x_biased = x + fit_side_bias[i_c]
y[i_c] = class_model.softmax(x_biased, v)
y_mean = np.zeros(len(x))
mean_side_bias = np.mean(fit_side_bias)
std_side_bias = np.std(fit_side_bias)
x_biased = x + mean_side_bias
y_mean[:] = class_model.softmax(x_biased, v_mean)
ax.axvline(0,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
x_label = r"$SEU(L_{right}) - SEU(L_{left})$"
y_label = "$P(Choose\,L_{right})$"
text = r'$\lambda=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$\n' \
+ r'$\gamma=' + f'{mean_side_bias:.2f}\pm{std_side_bias:.2f}' + '$\n'
elif class_model.__name__ == "AgentSide":
fit_precision = data['precision']
# fit_side_bias = data['side_bias']
x = np.linspace(-1, 1, n_points)
y = np.zeros((n_chunk, len(x)))
for i_c in range(n_chunk):
v = fit_precision[i_c]
y[i_c] = class_model.softmax(x, v)
y_mean = np.zeros(len(x))
y_mean[:] = class_model.softmax(x, v_mean)
ax.axvline(0,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
x_label = r"$SEU(L_{right}) - SEU(L_{left})$"
y_label = "$p(choose L_{right})$"
text = r'$\lambda=' + f'{v_mean:.2f}\pm{v_std:.2f}' + '$'
else:
raise ValueError
if linestyles is None:
linestyles = ['-' for _ in range(len(data['precision']))]
for i_c in range(n_chunk):
ax.plot(x,
y[i_c],
color=color,
linewidth=1,
alpha=0.5,
linestyle=linestyles[i_c])
# show_average
ax.plot(x, y_mean, color=color, linewidth=3, alpha=1)
add_text(ax, text)
# ax.set_xticks([0, 1, 2])
ax.set_yticks([0, 0.5, 1])
# ax.set_xlim(0.0, 2.01)
ax.set_ylim(-0.01, 1.01)
ax.spines['right'].set_color('none')
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.spines['top'].set_color('none')
ax.axhline(0.5,
alpha=0.5,
linewidth=1,
color='black',
linestyle='--',
zorder=-10)
ax.set_xlabel(x_label)
ax.set_ylabel(y_label)