Hi all?
Thank you very much for reading this email.
After trying the bigmemory package, I could not figure it out. So I turn to try
ff, hopefully someone here have some experience about it?
I have a data set (data.frame) with 16459 rows and 457 columns. I am
trying to do a multiple imputation for the missing values in the data
set. Usually, my ram (4G) is not enough for the imputation with such a big data,
so I figured maybe I could try
ff. However, I failed again, the ram was still not enough. Here is my code, is
there anything wrong with it?
> x=ff("acc3",vmode="double",dim=c(16459,457)) # acc3 is
the object
name of the data set
Warning messages:
1: In as.vmode.default(initdata[1], vmode) : NAs introduced by coercion
2: In as.vmode.default(value, vmode) : NAs introduced by
coercion> x=acc3
> fix(x)
> library(mice) # MICE package for multiple imputation for missing value.
Loading required package: MASS
Loading required package: nnet
Loading required package: lattice
mice 2.12 2012-03-25> acc3imp=mice(x,m=50,seed=1234,print=F) # m=50 means data set was
imputed for 50 times, thus, another 50 new
# data sets without missing value were generated.
Any advice are appreciated. Thank you very much.
Best regards,
YA