Classes derived from SplitDecisionObj. More...
Public Member Functions | |
SplitOnK (size_t k, size_t minCP=0) | |
bool | operator() () const |
bool | operator() (const SPSnode *const spn) const |
Classes derived from SplitDecisionObj.
These class provide functions provide a test for whether a node should be split when the histogram is being formed as each data point is inserted. The function will return true if the node should be split.
These classes are used with one-at-a-time insertion of data points into a subpaving tree.
Class for splitting based on number of points in a node.
This class will try to split a node with > maxPoints of data points associated with it, but will never split any node if the result would give at least one child with < minChildPoints of data associated with it. i.e. the class will only split a node if the node has > maxPoints of data and both children after the split will have >= minChildPoints of data.
Thus this class may leave a node with > maxPoints of data associated with it unsplit because splitting the node would create a child node with < minChildPoints of data associated with it.