In other words, the missing values are like a random sample of all the cases in the feature.
Since there exist no markers without missing values in the GBS-like marker data sets generated by random depletion, we instead first estimated RD matrices among pairs of genotypes based on the genotypic profiles by omitting for each pair of lines markers with missing values.
QIF – like standard GEE models – requires the assumption that missing values are "missing-completely-at-random" (MCAR) for complete case analysis [ 30].
Moreover imputing the missing values of continuous variables like ALT, AST and TLC with averages preserves the sample mean but the co-variance structure is distorted to the extent that estimates are biased towards zero.
We contrasted the above scenario with a scenario mimicking imputation of missing values for genotyping by sequencing (GBS -like data sets (Table 1).
The occurrence of missing values is not uncommon in real life databases like industrial, medical, and life science.
For other parameters like oocytes, embryos etc, if missing values occur, then they will not be imputed.
In other cases, like normalization of microarray data or imputation of missing values, however, this task is more complex.
A variety of analyses were performed, whereby options like 'fixed number of factors yes or no' and 'replace missing values by mean yes or no' were tested.
