hleapsApproxTime: Approximate hleaps runtime
In aasrinivasan/hleaps: Hierarchy Supporting Ordinary Least Squares

Description Usage Arguments Details Value Author(s) Examples

This method estimates the time needed to execute the hleaps algorithm for a given model size.

1	hleapsApproxTime(n=1000, p=10, int=TRUE, sigI=TRUE, sigC=TRUE)

`n`	an integer - the number of data observations in the data set
`p`	an integer - the number of variables to be used in subset selection. This count is only for the main effects.
`int`	a boolean - are interactions included in model? It TRUE, the estimate is for a model with all second order interaction terms. This provides an upper bound on the algorithm time.
`sigI`	a boolean - are interactions expected to be significant? If TRUE, the estimate is for a model with all the second order interaction terms being significant. This provides an upper bound on the algorithm time. False assumes none of the interaction terms are significant.
`sigC`	a boolean - are covariates expected to be significant? If TRUE, the estimate is for a model with all the main effects being significant. This provides an upper bound on the algorithm time. False assumes none of the main effects are significant.

The function prints out the estimated run time for the hleaps algorithm. This function times how long it takes to run hleaps on a calibrated data set on the currently running computer. This calibrated time is then scaled to the model using a formula provided below.

logTime = (7.135*(10^-7) * n + .01872 * p^2 + .0824 * p + -1.492* int + -2.252 * sigI + 1.563 * sigC - .28781) * (timeOnUserMachine)/(timeOnEstimationRuns)

This estimate is attained from running from data collected through timing estimation runs. The formula was generated from a regression on the results of multiple timing runs of hleaps with varying values for n, p, int, sigI and sig C. The linear regression was done on the log(times) collected for each run. A log normal correction was applied to the model.

This function will take at least 10 seconds to run in order to ensure a stable result.

An object of class list.

days - an integer. The approximate number of days needed for hleaps algorithm
hours - an integer. The approximate number of hours needed for hleaps algorithm
minutes - an integer. The approximate number of minutes needed for hleaps algorithm
seconds - an integer. The approximate number of seconds needed for hleaps algorithm

Arun Srinivasan <aasrinivasan@wisc.edu>