estimateWork {mspath} | R Documentation |
Compute estimates of the amount of work a calculator will need to do for each individual case. These may be much quicker to compute than the actual likelihood evaluation.
estimateWork(calc)
calc |
The calculator for which we want the work estimate.
Should be mspathCalculator or subclass. |
Subject to change. Various estimates of node and path counts, for the particular model and data currently in the calculator. I think this always does all cases, but need to check.
A matrix
whose rows are individual cases. The columns are
ID |
ID of the case for this row. |
Good Nodes |
Presumed to be the best single predictor of effort, in particular of floating-point operations. This is the number of unique nodes on good paths. |
Good Paths |
Number of distinct paths for this case, consistent with the model and the data. |
Bad Nodes |
Number of nodes considered that fell outside of the good paths. May vary with the pruning strategy used, but at any rate note that these nodes need not be constructed. |
Good Path Nodes |
This is the sum of all the nodes in all the good paths, ignoring the fact that some nodes are shared between paths. This is a measure of the work effort that a naive implementation would require. The current implementation is not naive. |
Steps |
The average number of steps on the good paths. If some paths can end before others, this value may not be an integer. |
The returned value is not stored in the calculator.
mspathDistributedCalculator uses this function to schedule work.
Ross Boylan