Lack of an 'auto' function

[Polkas]

Many users expecting less accurate but easier to implement solutions. auto_fill_NA could be a well suited proposition.

[Polkas]

When median<mean its means that there is a long right tail. Thus it might be preferred a log-transformation of the dependent variable before a regression.

[Polkas]

The fastest, simplest and clear solution will be to use bayes mean. For factor/character variables using a lda with continuous variables previously imputed.

[Polkas]

The mice package using pmm (for continuous and factors converted into integers) so it might be a better solution - partly. It could be well optimized in C++.