def run(self):
# get data
df_train = pd.read_csv(self.input()[0].path, header=[0, 1])
df_test = pd.read_csv(self.input()[1].path, header=[0, 1])
df_train_no_covariance_shift = pd.read_csv(self.input()[2].path,
header=[0, 1])
evaluation_setups = [EvaluationStruct("Proposed", rf)]
# evaluate the different methods
df_adapted = evaluate_data(df_train, noise_levels,
df_test, noise_levels,
evaluation_setups=evaluation_setups)
df_adapted["data"] = "adapted"
df_no_adaptation = evaluate_data(
df_train.drop("weights", axis=1), noise_levels,
df_test, noise_levels,
evaluation_setups=evaluation_setups)
df_no_adaptation["data"] = "source"
df_no_covariance_shift = evaluate_data(
df_train_no_covariance_shift, noise_levels,
df_test, noise_levels,
evaluation_setups=evaluation_setups)
df_no_covariance_shift["data"] = "target"
df = pd.concat([df_adapted, df_no_adaptation, df_no_covariance_shift])
# plot it
sns.boxplot(data=df, x="noise added [sigma %]", y="Errors", hue="data",
hue_order=["source", "adapted", "target"], fliersize=0)
# tidy up plot
plt.ylim((0, 40))
plt.legend(loc='upper left')
# finally save the figure
plt.savefig(self.output().path, dpi=500)
def run(self):
# get data
df_train = pd.read_csv(self.input()[0].path, header=[0, 1])
df_test = pd.read_csv(self.input()[1].path, header=[0, 1])
df = evaluate_data(df_train, noise_levels, df_test, noise_levels)
standard_plotting(df)
# finally save the figure
plt.savefig(self.output().path, dpi=500, bbox_inches='tight')
def run(self):
# get data
df_train = pd.read_csv(self.input()[0].path, header=[0, 1])
df_test = pd.read_csv(self.input()[1].path, header=[0, 1])
df = evaluate_data(df_train, noise_levels, df_test, noise_levels)
standard_plotting(df)
# finally save the figure
plt.savefig(self.output().path, dpi=500,
bbox_inches='tight')