CutAtBackgroundRejectionClassifier Class Reference

Inheritance diagram for CutAtBackgroundRejectionClassifier:

Public Member Functions
def	__init__ (self, background_rejection=0.5, cut_direction=1)
def	determine_cut_value (self, estimates, truths)

Public Attributes
	background_rejection
	cachec copy of the background-rejection threshold

Detailed Description

Apply cut on the background rejection

Definition at line 564 of file classification.py.

Constructor & Destructor Documentation

__init__()

def __init__	(		self,
			background_rejection = 0.5,
			cut_direction = 1
	)

Constructor

Reimplemented from CutClassifier.

Definition at line 567 of file classification.py.

    def __init__(self, background_rejection=0.5, cut_direction=1):
        """Constructor"""
        super().__init__(cut_direction=cut_direction, cut_value=np.nan)
        
        self.background_rejection = background_rejection
 

Member Function Documentation

determine_cut_value()

def determine_cut_value	(		self,
			estimates,
			truths
	)

Find the cut value that satisfies the desired background-rejection level

Reimplemented from CutClassifier.

Definition at line 573 of file classification.py.

    def determine_cut_value(self, estimates, truths):
        """Find the cut value that satisfies the desired background-rejection level"""
        n_data = len(estimates)
        n_signals = scores.signal_amount(truths, estimates)
        n_bkgs = n_data - n_signals
 
        sorting_indices = np.argsort(estimates)
        if self.cut_direction_ < 0:  # reject low
            # Keep a reference to keep the content alive
            original_sorting_indices = sorting_indices  # noqa
            sorting_indices = sorting_indices[::-1]
 
        sorted_truths = truths[sorting_indices]
        sorted_estimates = estimates[sorting_indices]
 
        sorted_n_accepted_signals = np.cumsum(sorted_truths, dtype=float)
        # sorted_efficiencies = sorted_n_accepted_signals / n_signals
 
        sorted_n_rejected_signals = n_signals - sorted_n_accepted_signals
        sorted_n_rejects = np.arange(len(estimates) + 1, 1, -1)
        sorted_n_rejected_bkgs = sorted_n_rejects - sorted_n_rejected_signals
        sorted_bkg_rejections = sorted_n_rejected_bkgs / n_bkgs
 
        cut_index, = np.searchsorted(sorted_bkg_rejections[::-1], (self.background_rejection,), side='right')
 
        cut_value = sorted_estimates[-cut_index - 1]
        return cut_value

Member Data Documentation

background_rejection

background_rejection

cachec copy of the background-rejection threshold

Definition at line 571 of file classification.py.

---

The documentation for this class was generated from the following file:

• tracking/scripts/tracking/validation/classification.py