r - Omit numerous NA values -

this question has answer here:

• partially merge 2 datasets , fill in nas in r 1 answer

i have df like:

a  b  c  d | va na 2  na 3 | x 1  2  3  4 | x na 5  5  2 | x  4  3  2  1 | x 

it needed like:

a  b  c  d | va 1  2  3  4 | x 4  3  2  1 | x 

the aim of calculate va value. have calculate rules (lets name vb), has low quality. therefore, need calculate va better values, can (where there no na in row). need v vector, va calculated rows , vb everywhere else. illustration of need:

va  vb  ->  v na  2   ->  2 1   2   ->  1 na  2   ->  2 1   2   ->  1 

so, have 2 questions:

1) how rescale original df newdf, each row has no nas respect original index in df?

2) have tried va[is.na(va)] <- vb test na replacement vb values (low quality model). worked message different length/number of replacements. ok use need or maybe there better?

we can use rowsums , is.na subset rows of "df"

df[!rowsums(is.na(df)),] #   b c d va #2 1 2 3 4  x #4 4 3 2 1  x