format="pdf", dpi=800)
fig.savefig(os.path.join(baseline_working_directory, "Output", "ICML18_Figure5_Nile.png"),
format="png", dpi=800)
# Check that the figure has indeed been generated and report
if os.path.isfile(os.path.join(baseline_working_directory, "Output", "ICML18_Figure5_Nile.pdf")):
print("Figure 5 can be found in the Output directory")
"""STEP 4: Generate additional figures, if requested"""
if save_extra_figures:
# Plot the prediction error and variance
fig_pe, ax_pe = plt.subplots()
EvT.plot_prediction_error(river_height.reshape(T, 1, 1), indices=[0],
show_var=True, show_MAP_CPs=True,
up_to=250, ax=ax_pe)
fig_pe.savefig(os.path.join(baseline_working_directory, "Output", "ICML18_ExtraFigure_Nile_PredError.pdf"),
format="pdf", dpi=800)
if os.path.isfile(os.path.join(baseline_working_directory, "Output", "ICML18_ExtraFigure_Nile_PredError.pdf")):
print("Extra plot of the prediction error can be found in the Output directory")
# Plot the model posterior
fig_mp, ax_mp = plt.subplots()
EvT.plot_model_posterior(indices=[0, 1, 2], log_format=False, up_to=150,
plot_type="MAPVariance1_det", show_MAP_CPs=True, ax=ax_mp)
fig_mp.savefig(os.path.join(baseline_working_directory, "Output", "ICML18_ExtraFigure_Nile_ModelPosterior.pdf"),
format="pdf", dpi=800)
if os.path.isfile(os.path.join(baseline_working_directory, "Output", "ICML18_ExtraFigure_Nile_ModelPosterior.pdf")):
print("Extra plot of the model posterior can be found in the Output directory")
fig, ax_array = plt.subplots(num_subplots, sharex = True,
gridspec_kw = {'height_ratios':height_ratio})#,
#figsize = (12,5))
plt.subplots_adjust(hspace = .185, left = None, bottom = None, right = None, top = None)
"""plot raw data"""
fig_1 = EvT.plot_raw_TS(data, indices=[0,2,4],
custom_colors_series = custom_colors_series,
ax = ax_array[0],xlab=None,
ylabel_coords = ylabel_coords) #ax_array[0])
"""Plot + save pics (4) prediction + error"""
fig_2 = EvT.plot_prediction_error(show_var= True, data = data_original,
indices = [0],
custom_colors = ['black', 'darkgray'],
ax=ax_array[1],
time_range = np.linspace(1,T-2,T-2,dtype=int),
aspect_ratio = 'auto', xlab = None, ylab = "PE",
ylabel_coords = ylabel_coords)
"""Plot + save pictures: (3) model posteriors"""
"""Plot + save pictures: (2) Raw TS with MAP CPs"""
#backconverter = tick/C2-C1 (backconv+C1)*C2 = tick
C1, C2 = 100, 1
#raw_ticks = [pow(10, -18), pow(10, -15), pow(10, -13), pow(10, -11)]
colorbar_ticks = None #[C2*(r+C1) for r in raw_ticks]
fig_4 = EvT.plot_run_length_distr(mark_max=True,upper_limit=250,
aspect_ratio='auto', C1=C1, C2=C2,
time_range = np.linspace(1,T-2,T-2, dtype=int),
CP_legend = True, CP_legend_fontsize = 8,
CP_custom_legend_labels = ["AR(2)", "AR(3)", "VAR4(2)"],
trim_type="keep_K",
threshold=200,
training_period=25, #i.e., we let 2 years pass before MSE computed
notifications=1500,
save_performance_indicators=True)
detector.run()
"""Store results + real CPs into EvaluationTool obj"""
EvT = EvaluationTool()
EvT.build_EvaluationTool_via_run_detector(detector)
"""plot transformed data"""
fig = EvT.plot_raw_TS(data.reshape(T, 1))
"""plot prediction error"""
fig = EvT.plot_prediction_error(data,
indices=[0],
print_plt=True,
time_range=np.linspace(2 * 365,
T - upper_AR - 1,
T - upper_AR - 1 -
2 * 365,
dtype=int))
"""plot predictions themselves"""
fig = EvT.plot_predictions(indices=[0],
print_plt=True,
legend=False,
legend_labels=None,
legend_position=None,
time_range=None,
show_var=False,
show_CPs=True)
plt.show()