| rake {rake} | R Documentation |
A rake class object contains sample weights for a sample
dataset,
broken down by the factors of the two categorical response variables
specified by
colCol and rowCol.
This is the first function in the rake, rakeadj,
predict.rake Rake analysis series.
rake(data,
colCol = 1,
rowCol = 2,
weight = if("weight" %in% names(data)) "weight" else 1 )
data |
data.frame of sample observations. |
colCol |
column of data whose data will be put in the
columns of the rake object. |
rowCol |
column of data whose data will be put in the
rows of the rake object. |
weight |
weights of sample observations, specified as either the
column name of data that contains the weights, or a numeric
vector of weights that will be replicated to be as long as
nrow(data). |
rake treats the columns of data specified by
colCol and rowCol as factors and constructs a matrix
that contains sample weights.
The rake class object has
format, predict, and print S3 methods.
Raking is the 2D analog of weighting-class adjustment. See References.
Object of class rake that contains sample weights broken down
by class.
Toby Dylan Hocking <tdhock@ocf.berkeley.edu>
Sharon L. Lohr. Sampling: Design and Analysis, pp. 269-271. Brooks/Cole, Pacific Grove, CA, 1999.
rakeadj for performing the rake adjustment,
predict.rake for making estimates with the adjusted weight
values,
simpleRake for performing the entire Raking process.
data(certify)
# Merge some classes since we have limited population data
# We define classes of COLLEGE as P or not P = N
certify$COLLEGE[ certify$COLLEGE != "P" ] <- "N"
# We define classes of WORKENV to be I, A, or else = O
certify$WORKENV[ ! certify$WORKENV %in% c("I","A") ] <- "O"
r <- rake( certify, "COLLEGE", "WORKENV" )
print(r) # Uses print.rake