def plot_sudden_markov_multinutr_environments(input_fname, plot_fname, label):
"""
Plot sudden multinutrient environments.
"""
print "plotting: %s" %(os.path.basename(plot_fname))
# set seed
np.random.seed(2)
plt.figure(figsize=(4.5, 4.05))
# transition parameters
trans_params = [np.array([[0.9, 0.05, 0.05],
[0.05, 0.9, 0.05],
[0.05, 0.05, 0.9]]),
np.array([[0.05, 0.9, 0.05],
[0.9, 0.05, 0.05],
[0.05, 0.05, 0.9]]),
np.array([[0.05, 0.9, 0.05],
[0.1, 0.6, 0.3],
[0.05, 0.05, 0.9]])]
# initial state probabilities: start with Glu
init_probs = np.array([1., 0., 0.])
time_obj = time_unit.Time(0, 100, step_size=1)
num_points = len(time_obj.t)
#plt.axvspan(a, b, color='y', alpha=0.5, lw=0)
num_plots = len(trans_params)
n = 0
data_to_labels = {0: "Glu",
1: "Gal",
2: "Mal"}
labels_to_colors = \
{"Glu": sns.color_palette("Set1")[2],
"Gal": sns.color_palette("Set1")[1],
"Mal": plot_utils.orange}
sns.set_style("ticks")
num_xticks = 11
palette = "seismic"
prev_n = 0
prev_k = 0
gs = gridspec.GridSpec(num_plots * 3, 3)
gs.update(hspace=0.6)
# order of nutrients along transition matrix
nutr_order = ["Glu", "Gal", "Mal"]
for n in xrange(num_plots):
trans_mat = trans_params[n]
trans_df = pandas.DataFrame(trans_mat)
trans_df.columns = nutr_order
trans_df.index = nutr_order
ax1 = plt.subplot(gs[prev_n:prev_n+3, 0])
heatmap_ax = sns.heatmap(trans_df, annot=True, cbar=False,
cmap=plt.cm.gray_r,
linewidth=0.1,
linecolor="k",
annot_kws={"fontsize": 8})
heatmap_ax.set_aspect("equal")
ax1.tick_params(axis="both", which="major", pad=0.01, length=2.5,
labelsize=8)
if n != (num_plots - 1):
plt.axis("off")
ax2 = plt.subplot(gs[prev_k+1:prev_k + 2, 1:3])
prev_n += 3
prev_k += 3
# sample values from current transition matrix
samples = prob_utils.sample_markov_chain(num_points,
init_probs,
trans_mat)
handles_info = \
plot_utils.plot_sudden_switches(time_obj, samples,
data_to_labels=data_to_labels,
labels_to_colors=labels_to_colors,
ax=ax2,
pad=0.01,
despine=True)
if n == (num_plots - 1):
plt.xlabel("Time step", fontsize=8)
plt.savefig(plot_fname)
def plot_sudden_ssm_filtering(input_fname, plot_fname, label):
"""
Plot SSM filtering predictions.
"""
print "plotting: %s" %(os.path.basename(plot_fname))
fig = plt.figure(figsize=(7, 5))
sns.set_style("ticks")
all_results = simulation.load_data(input_fname)
all_results = all_results["model"]
num_plots = len(all_results)
total_plots = num_plots * 2
gs = gridspec.GridSpec(total_plots, 1,
height_ratios=[1, 0.2]*num_plots)
curr_plot_num = 0
axes = {}
for n, data_label in enumerate(all_results.keys()):
data_set = all_results[data_label]
params = data_set["params"]
time_obj = time_unit.Time(params["t_start"],
params["t_end"],
step_size=params["step_size"])
c = 0.8
x_axis = time_obj.t[0::4]
xlims = [time_obj.t[0] - c, time_obj.t[-1] + c]
ax1 = plt.subplot(gs[curr_plot_num, 0])
pred_probs = [p[0] for p in data_set["preds_with_lag"]]
plt.plot(time_obj.t, pred_probs, "-o", color=plot_utils.red,
label="Prediction",
clip_on=False,
zorder=100)
plt.xlabel(r"Time step")
plt.ylabel(r"$P(C_{t+1} =\ \mathsf{Glu} \mid C_{0:t})$",
fontsize=11)
plt.title("lag = %d" %(params["decision_lag_time"]), fontsize=8)
plt.legend(loc="lower right")
plt.xticks(x_axis, fontsize=8)
plt.xlim(xlims)
plt.ylim([0, 1])
plt.yticks(np.arange(0, 1 + 0.2, 0.2))
ax2 = plt.subplot(gs[curr_plot_num + 1, 0])
data_to_labels = {0: "Glu", 1: "Gal"}
labels_to_colors = {"Glu": plot_utils.green, "Gal": plot_utils.blue}
data = params["data"]
ax2.get_yaxis().set_visible(False)
ax2.set_yticks([])
ax2.spines["left"].set_visible(False)
plot_utils.plot_sudden_switches(time_obj, data,
data_to_labels=data_to_labels,
labels_to_colors=labels_to_colors,
box_height=0.025,
y_val=0.02,
ax=ax2,
despine=False,
with_legend=True,
legend_outside=(0,1))
plt.xticks(x_axis, fontsize=8)
plt.xlim(xlims)
# despine axes
sns.despine(trim=True, left=True, ax=ax2)
sns.despine(trim=True, ax=ax1)
# advance number of plots by two
curr_plot_num += 2
ax1.spines["left"].set_visible(True)
#plt.tight_layout(h_pad=0.1)
fig.set_tight_layout(True)
plt.savefig(plot_fname)
