Model

Model: grouping classifiers for different histogram types

This class represents a general model for assigning a score to a lumisection.
It consists of two distinct parts:

• a collection of classifiers acting on individual histogramgs (one for each type).
• a fitter to assign a probability density to the output scores obtained in the previous step.
  The types of histograms, classifiers, and fitter can be freely chosen.

This class does not contain the histograms or other data; it only contains the classifiers and fitter.
Use the derived class ModelInterface to make the bridge between a HistStruct (holding the data)
and a Model (holding the classifiers and fitter).

---

---

[class] Model

comments:

(no valid documentation found)  

⤷ __init__

full signature:

def __init__( self, histnames )  

comments:

initializer  
input arguments:  
- histnames: list of the histogram names for this Model.  
  this is the only argument needed for initialization,  
  use the functions set_classifiers and set_fitter   
  to set the classifiers and fitter respectively.  

⤷ set_classifiers

full signature:

def set_classifiers( self, classifiers )  

comments:

set the classifiers for this Model  
input arguments:  
- classifiers: dict of histnames to classifiers.  
  the histnames must match the ones used to initialize this Model,  
  the classifiers must be a subtype of HistogramClassifier.  

⤷ set_fitter

full signature:

def set_fitter( self, fitter )  

comments:

set the fitter for this Model  
input arguments:  
- fitter: an (untrained) object of type CloudFitter  

⤷ check_classifier

full signature:

def check_classifier( self, histname )  

comments:

check if a classifier was initialized  
input arguments:  
- histname: type of histogram for which to check the classifier  

⤷ check_fitter

full signature:

def check_fitter( self )  

comments:

check if a fitter was initialized  

⤷ train_classifier

full signature:

def train_classifier( self, histname, histograms, **kwargs )  

comments:

train a classifier  
input arguments:  
- histname: histogram name for which to train the classifier  
- histograms: the histograms for training, np array of shape (nhistograms,nbins)  
- kwargs: additional keyword arguments for training  

⤷ train_classifiers

full signature:

def train_classifiers( self, histograms, **kwargs )  

comments:

train classifiers for all histograms in this Model  
input arguments:  
- histograms: the histograms for training, dict of histnames to np arrays of shape (nhistograms,nbins)  
- kwargs: additional keyword arguments for training  

⤷ evaluate_classifier

full signature:

def evaluate_classifier( self, histname, histograms, mask=None )  

comments:

evaluate a classifier and return the score  
input arguments:  
- histname: histogram name for which to evaluate the classifier  
- histograms: the histograms for evaluation, np array of shape (nhistograms,nbins)  
- mask: a np boolean array masking the histograms to be evaluated  
returns:  
- a np array of shape (nhistograms) with the scores  
note: masked-out indices are set to np.nan!  

⤷ evaluate_classifiers

full signature:

def evaluate_classifiers( self, histograms, mask=None )  

comments:

evaluate the classifiers and return the scores  
input arguments:  
- histograms: dict of histnames to histogram arrays (shape (nhistograms,nbins))  
- mask: a np boolean array masking the histograms to be evaluated  
returns:  
- dict of histnames to scores (shape (nhistograms))  
note: masked-out indices are set to np.nan!  

⤷ get_point_array

full signature:

def get_point_array( self, points )  

comments:

for internal use in train_fitter and evaluate_fitter  
input arguments:  
- points: dict matching histnames to scores (np array of shape (nhistograms))  

⤷ train_fitter

full signature:

def train_fitter( self, points, verbose=False, **kwargs )  

comments:

train the fitter  
input arguments:  
- points: dict matching histnames to scores (np array of shape (nhistograms))  
- kwargs: additional keyword arguments for fitting  

⤷ evaluate_fitter

full signature:

def evaluate_fitter( self, points, mask=None, verbose=False )  

comments:

evaluate the fitter and return the scores  
input arguments:  
- points: dict matching histnames to scores (np array of shape (nhistograms))  
- mask: a np boolean array masking the histograms to be evaluated  
returns:  
- a np array of shape (nhistograms) with the scores  
note: masked-out indices are set to np.nan!  

---