def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[str] = None) -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = df_in.copy()
df['average_return'] = df.raw_return.diff() / df.episode.diff()
p = plotting.plot_individual_returns(
df_in=df[df.episode > 1],
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var='height_threshold',
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[str] = None) -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = dc_preprocess(df_in)
df['average_return'] = 1.1 - (df.total_regret.diff() / df.episode.diff())
p = plotting.plot_individual_returns(
df_in=df,
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var='optimal_horizon',
yintercept=1.1,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[Text] = None,
colour_var: Text = 'memory_length') -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = df_in.copy()
df['average_return'] = df.total_return.diff() / df.episode.diff()
p = plotting.plot_individual_returns(
df_in=df[df.episode > 10],
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var=colour_var,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[str] = None,
colour_var: str = None) -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = df_in.copy()
df['average_return'] = df.raw_return.diff() / df.episode.diff()
p = plotting.plot_individual_returns(
df_in=df,
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var=colour_var,
yintercept=-_SOLVED_STEPS,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Optional[Sequence[str]] = None,
colour_var: Optional[str] = None) -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = df_in.copy()
df['average_return'] = 1.0 - (df.total_regret.diff() / df.episode.diff())
p = plotting.plot_individual_returns(
df_in=df,
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var=colour_var,
yintercept=1.,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[Text] = None,
yintercept: float = 0.99) -> gg.ggplot:
"""Plot the returns through time individually by run."""
df = df_in.copy()
df['average_return'] = df.denoised_return.diff() / df.episode.diff()
p = plotting.plot_individual_returns(
df_in=df[df.episode > 0.01 *
NUM_EPISODES], # First episodes very noisy
max_episode=NUM_EPISODES,
return_column='average_return',
colour_var='size',
yintercept=yintercept,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average episodic return')
def plot_seeds(df_in: pd.DataFrame,
sweep_vars: Sequence[Text] = None,
colour_var: Text = None) -> gg.ggplot:
"""Plot the accuracy through time individually by run."""
df = df_in.copy()
df['average_return'] = 1.0 - (df.total_regret.diff() / df.episode.diff())
df['average_accuracy'] = (df.average_return + 1) / 2
p = plotting.plot_individual_returns(
df_in=df[df.episode >= 100],
max_episode=NUM_EPISODES,
return_column='average_accuracy',
colour_var=colour_var,
yintercept=1.,
sweep_vars=sweep_vars,
)
return p + gg.ylab('average accuracy')
