R help - Dec 2015 - Weighted demean by group on only a selection of the dataset

Dear colleagues,

I am trying to find a simple code to demean
1) only certain values of a dataset,
2) by group
3) and in a weighted fasion.

Currently, I can only demean all the numeric variables in the dataset:

Data[,sapply(Data,  is.numeric)] <- apply(Data[sapply(Data,
is.numeric)], 2, function(x) scale(x, scale = FALSE))

Assume that my dataset looks like this:
Country<-
c('BE','BE','DE','GR','IT','ES','DE','NL')
Landvalue<- c(21000, 23400, 26800, 15000,18000,23000,19000,23000)
Temperature_spring <-
c('15','16','14','18','23','21','12','15')
Temperature_summer <-
c('25','18','19','23','24','22','15','19')
Temperature_autumn <-
c('14','12','12','10','20','20','11','13')
Temperature_winter <-
c('9','4','12','14','15','13','17','12')
Weight<-c('5','20','3','2','15','21','13','8')
Data <- data.frame(Country, Landvalue,
Temperature_spring,Temperature_summer,
Temperature_autumn,Temperature_winter, Weight)


Now imagine I only want to demean the temperature-variables, grouped
by country and weighted by weight. With grouped by country I mean that
I want to subtract only the mean of Belgium from an observation in
Belgium.

Does somebody know how to add the three functions to the code line I
already have? Or if this does not work, what code should I use?

Thank you very much and have a nice Christmas!

> On Dec 24, 2015, at 5:03 AM, Janka VANSCHOENWINKEL
<janka.vanschoenwinkel at uhasselt.be> wrote:
> 
> Dear colleagues,
> 
> I am trying to find a simple code to demean
> 1) only certain values of a dataset,
> 2) by group
> 3) and in a weighted fasion.
> 
> Currently, I can only demean all the numeric variables in the dataset:
> 
> Data[,sapply(Data,  is.numeric)] <- apply(Data[sapply(Data,
> is.numeric)], 2, function(x) scale(x, scale = FALSE))
> 
> Assume that my dataset looks like this:
> Country<-
c('BE','BE','DE','GR','IT','ES','DE','NL')
> Landvalue<- c(21000, 23400, 26800, 15000,18000,23000,19000,23000)
> Temperature_spring <-
c('15','16','14','18','23','21','12','15')
> Temperature_summer <-
c('25','18','19','23','24','22','15','19')
> Temperature_autumn <-
c('14','12','12','10','20','20','11','13')
> Temperature_winter <-
c('9','4','12','14','15','13','17','12')
>
Weight<-c('5','20','3','2','15','21','13','8')
> Data <- data.frame(Country, Landvalue,
> Temperature_spring,Temperature_summer,
> Temperature_autumn,Temperature_winter, Weight)
Do note that only the `Landvalue` column is numeric above. You would need to us
as.numeric on the vectors that you are nominating for processing below.
> 
> 
> Now imagine I only want to demean the temperature-variables, grouped
> by country and weighted by weight. With grouped by country I mean that
> I want to subtract only the mean of Belgium from an observation in
> Belgium.
> 
Generally when one wants to use multiple columns in a calculation with grouping,
the method needs to be along the lines of :

Data[ , grepl("Temp", Data) ] <- 
       lapply( split(Data, Data$Country), FUN= ...) 

> Does somebody know how to add the three functions to the code line I
> already have? Or if this does not work, what code should I use?
> 
> Thank you very much and have a nice Christmas!
> 
> 
-- 

David Winsemius
Alameda, CA, USA