Ctrl+K

gnn_tracking.analysis.efficiencies

gnn_tracking.analysis.efficiencies#

Module Contents#

Classes#

TracksVsDBSCANPlot

Plot tracking efficiencies vs DBSCAN epsilon

PerformancePlot

Plot efficiencies vs some variable (pt, eta, etc.)

PerformanceComparisonPlot

Similar to PerforamncePlot, except that we use the same x axis for
class gnn_tracking.analysis.efficiencies.TracksVsDBSCANPlot(mean_df: pandas.DataFrame, **kwargs)#

Bases: gnn_tracking.analysis.plotutils.Plot

Plot tracking efficiencies vs DBSCAN epsilon

tvdp = TracksVsDBSCANPlot(
    mean_df=tcmodule.cluster_scanner.get_results().df_mean,
)
secondary_k = 4
tvdp.plot_var("double_majority_pt0.9", secondary_k=secondary_k)
tvdp.plot_var("lhc_pt0.9", secondary_k=secondary_k)
tvdp.plot_var("perfect_pt0.9", secondary_k=secondary_k)
plot_var(var: str, *, secondary_k: int = 4, **kwargs)#

Plot an efficiency.

Parameters:

  • var – Name of the variable to plot

  • secondary_k – Plot a second line with this value of k

  • **kwargs – Passed to plot function

class gnn_tracking.analysis.efficiencies.PerformancePlot(xs: numpy.ndarray, df: pandas.DataFrame, *, df_ul: pandas.DataFrame | None = None, x_label: str = '$p_T$ [GeV]', y_label: str = 'Efficiency', **kwargs)#

Bases: gnn_tracking.analysis.plotutils.Plot

Plot efficiencies vs some variable (pt, eta, etc.)

Parameters:

  • xs (np.ndarray) – x values (e.g., pt or eta). Length must be one longer than the dataframe to account for bin edges.

  • df (pd.DataFrame) – Dataframe with values. Errors should be in columns named with suffix _err.

  • df_ul (_type_, optional) – Dataframe with values for upper limit. Defaults to None.

  • x_label (regexp, optional) – x label

  • y_label (str, optional) – y label

  • **kwargs – Passed to Plot

plot_var(var: str, color: str, *, label: str | None = None, plot_ul=True) → None#

Plot variable

Parameters:

  • var (str) – Name of variable

  • color (str) – Color

  • label (str | None, optional) – Label for legend

  • plot_ul (bool, optional) – Plot upper limit if available

add_blocked(a: float, b: float, label='Not trained for') → None#

Used to mark low pt as “not trained for”.

add_legend(**kwargs) → None#
class gnn_tracking.analysis.efficiencies.PerformanceComparisonPlot(xs: numpy.ndarray, var: str, x_label: str, ylabel: str = 'Efficiency', **kwargs)#

Bases: gnn_tracking.analysis.plotutils.Plot

Similar to PerforamncePlot, except that we use the same x axis for plots of different models (and supply the dataframes directly to plot_var).

Parameters:

  • xs (np.ndarray) – x values (e.g., pt or eta). Length must be one longer than the dataframe to account for bin edges.

  • var (str) – Name of variable

  • x_label (regexp, optional) – x label

  • y_label (str, optional) – y label

  • **kwargs – Passed to Plot

plot_var(df: pandas.DataFrame, label: str, color: str) → None#
add_legend(**kwargs) → None#
add_blocked(a, b, label='Not trained for') → None#

Used to mark low pt as “not trained for”.

Contents