datasetsnames = np.unique(df_beam.datasetname)
results2plot = dict()
for datname in datasetsnames:
results2plot[datname] = dict()
results2plot[datname]["beamsize"] = df_beam[df_beam.datasetname ==
datname].beam_width.to_numpy()
results2plot[datname]["compression"] = df_beam[
df_beam.datasetname == datname].length_ratio.to_numpy()
results2plot[datname]["wkl_sum"] = df_beam[df_beam.datasetname ==
datname].wkl_sum.to_numpy()
fig, lgd = make_graph(results2plot,
"beamsize",
"compression",
size_marker=6,
color=tableau20,
typeofplot="semilogx",
separate_colour=1)
plt.axvline(x=100, linestyle='--', linewidth=0.6, color='k')
plt.xlabel("beam width")
plt.ylabel("relative compression")
save_path = os.path.join(folder_path, "beam_vs_compression.pdf")
fig.savefig(save_path, bbox_extra_artists=(lgd, ), bbox_inches='tight')
###############################################################################
# beamsize numeric targets
###############################################################################
folder_load = os.path.join("results", "hyperparameter testing",
"gaussian_beam_width_results", "summary.csv")
results2plot = dict()
for datname in datasetsnames:
results2plot[datname] = dict()
results2plot[datname]["maxdepth"] = df[df.datasetname ==
datname].maxdepth.to_numpy()
results2plot[datname]["compression"] = df[df.datasetname ==
datname].length_ratio.to_numpy()
results2plot[datname]["time"] = df[df.datasetname ==
datname].runtime.to_numpy()
results2plot[datname]["conditions"] = df[df.datasetname ==
datname].avg_items.to_numpy()
fig, lgd = make_graph(results2plot,
"maxdepth",
"compression",
size_marker=6,
color=tableau20,
typeofplot="plot",
separate_colour=1)
plt.axvline(x=5, linestyle='--', linewidth=0.6, color='k')
plt.xlabel("maximum depth")
plt.ylabel("relative compression")
save_path = os.path.join(folder_path, "maxdepth_vs_compression.pdf")
fig.savefig(save_path, bbox_extra_artists=(lgd, ), bbox_inches='tight')
fig, lgd = make_graph(results2plot,
"maxdepth",
"time",
size_marker=6,
color=tableau20,
typeofplot="semilogy",
df_ncut = pd.read_csv(folder_load, index_col=False)
datasetsnames = np.unique(df_ncut.datasetname)
results2plot = dict()
for datname in datasetsnames:
results2plot[datname] = dict()
results2plot[datname]["ncutpoints"] = df_ncut[
df_ncut.datasetname == datname].n_cutpoints.to_numpy()
results2plot[datname]["compression"] = df_ncut[
df_ncut.datasetname == datname].length_ratio.to_numpy()
#results2plot[datname]["time"] = df_ncut[df_ncut.datasetname == datname].runtime.to_numpy()
fig, lgd = make_graph(results2plot,
"ncutpoints",
"compression",
size_marker=6,
color=tableau20,
typeofplot="plot",
separate_colour=1)
plt.axvline(x=5, linestyle='--', linewidth=0.6, color='k')
plt.xlabel("number cutpoints")
plt.ylabel("relative compression")
save_path = os.path.join(folder_path, "ncutpoints_vs_compression.pdf")
fig.savefig(save_path, bbox_extra_artists=(lgd, ), bbox_inches='tight')
"""
fig,lgd = make_graph(results2plot,"ncutpoints","time",size_marker = 6,color = tableau20,typeofplot ="semilogy",separate_colour =1)
plt.axvline(x=5,linestyle= '--', linewidth=0.6, color='k')
plt.xlabel("number cutpoints")
plt.ylabel("rumtime (s)")
save_path = os.path.join(folder_path,"ncutpoints_vs_time.pdf")
fig.savefig(save_path, bbox_extra_artists=(lgd,), bbox_inches='tight')