tuneM: Tune the Number of Imputations in MI-MFA

In GonzalezIgnacio/HandlingMissRows: Handling Missing Individuals in Multi-Omics Data Integration

Description

tuneM can be used to determine the appropriate number of imputed datasets needed to obtain satisfactory results with MI-MFA.

Usage

tuneM(object, ncomp=2, Mmax=30, inc=5, N=10, tol=1e-06, showPlot=TRUE)

## S3 method for class 'tuneM'
print(x, ...)

Arguments

object	an object of class MIDTList.
x	an object of class inheriting from tuneM.
ncomp	a number of components to include in MFA.
Mmax	an integer corresponding to the maximum number of imputed datasets. See Details.
inc	integer. The increment of the sequence for the number M of imputations considered. See Details.
N	integer. Collections of size N are generated for each number of imputations M. See Details.
tol	a positive value, the tolerance used for assessing stabilization.
showPlot	logical. If TRUE (the default) a plot showing the stability of the estimated MFA configurations is displayed.
...	not currently used.

Details

The appropriate number of imputations can be informally determined by carrying out MI-MFA on N replicate sets of M_l imputations for l = 0, 1, 2, . . . , with M_0 < M_1 < M_2 < . . . < M_max, until the estimate compromise configurations are stabilized.

tuneM function implements such a procedure. Collections of size N are generated for each number of imputations M, with M = seq(inc, Mmax, by = inc). The stability of the estimated MI-MFA configurations is then determined by calculating the RV coefficient between the configurations obtained using M_l and M_{l+1} imputations.

If showPlot = TRUE a plot showing the stability of the estimated MFA configurations is displayed. The values shown are the mean RV coefficients for the N configurations as a function of the number of imputations. Error bars represent the standard deviation of the RV coefficients.

Value

A list with the following components:

stats	a data.frame containing the information used to generate the plot.
ggp	an object of class ggplot.

Author(s)

Ignacio González, Valentin Voillet

References

Voillet V., Besse P., Liaubet L., Cristobal M.S., González I. (2016). Handling missing rows in multi-omics data integration: Multiple Imputation in Multiple Factor Analysis framework. BMC Bioinformatics, 17(40).

See Also

MIMFA

Examples

#-- load data and create MIDTList object
data(NCI60)
midt <- MIDTList(NCI60$mae)

#-- tune the number of imputations
## Not run: 
tune <- tuneM(midt, ncomp=2, Mmax=100, inc=10, N=10)
tune
## End(Not run)

GonzalezIgnacio/HandlingMissRows documentation built on Jan. 17, 2020, 6:29 p.m.