#include <EvidenceType.h>
Public Member Functions | |
EvidenceType (const string &str, int idx, double val) | |
virtual void | update (const Prediction *pred, int end5, int end3, dsu::Strand_t)=0 |
takes a Prediction about a given region of the data where that Prediction correspons to a particular evidence type, and updates it's values in the feature vector | |
virtual void | finish () |
virtual unsigned | getNum () const=0 |
virtual double | getVal (int) const |
returns the i th feature value some features might be encoded with dependent values such as homology data that stores best match and total sum of matches | |
virtual EvidenceType * | dup () const=0 |
create a duplicate of this evidence type used to create heterogeneous feature vectors without having to actually know what type of evidence the feature vector is made up of | |
virtual void | setVals (const EvidenceType &et) |
set vals | |
virtual AnnotationScore | eval () const=0 |
virtual bool | train_print (std::ostream &os) const=0 |
during the training mode, method defines how this element of the feature vector should look (in print) | |
const char * | getStr () const |
get the string that identifies this type of evidence | |
const int | getIdx () const |
get a unique integer index that identifies this type of evidence | |
virtual bool | read (std::istream &)=0 |
double & | getValRef () |
virtual double | getVal () const |
virtual double | getSumVal () const |
virtual void | normalize (double sum) |
Protected Attributes | |
double | _val |
Note that the instanciated derived types of this class, must correspond to an element in the Prediction class heirarchy. In other words types of Prediction objects map to types of evaluation feature objects (there is not an explicit relationship established in the code)
|
set vals
|
|
takes a Prediction about a given region of the data where that Prediction correspons to a particular evidence type, and updates it's values in the feature vector
|