sqlrepsurvey: Large survey with replicate weights, based on SQL database
In sqlsurvey: analysis of very large complex survey samples (experimental)

Description Usage Arguments Details Value See Also Examples

Specifies a survey design with replicate weights based on a connection to a relational database (currently MonetDB).

sqlrepsurvey(weights, repweights, scale, rscales, driver=MonetDB.R(), database, table.name, key = "row_names", mse = FALSE, check.factors = 10, degf=NULL,...)
## S3 method for class 'sqlrepsurvey'
close(con, ...)
## S3 method for class 'sqlrepsurvey'
open(con, driver,...)

`weights`	Character string naming the weight variable
`repweights`	Vector of character strings naming the replicate-weight variables, or a regular expression that selects the correct variable names from those in the table.
`scale`	A single number for scaling the sum of squared deviations of the replicates.
`rscales`	A vector of the same length as `repweights` giving per-replicate scaling for squared deviations of the replicates
`driver`	A database driver object (eg returned by `MonetDB()`) or `NULL` if the `database` argument is already a database connection.
`database`	Either a connection to a MonetDB database or a character string with the name (URL) of a database containing the data table
`table.name`	A character string with the name of data table containing the data and replicate weights
`key`	A character string with the name of a unique identified variable.
`mse`	If `TRUE` compute standard errors based on deviations of replicates from the point estimate, if `FALSE` use deviations from the mean of the replicates.
`check.factors`	If this is a non-zero number, R will attempt to determine which variables in the database table are factors based on having at most this many distinct values, and will store information on the levels in the survey design object. This can be slow for a very large survey. `check.factors` can also be a zero-row data frame with the correct factor levels for the factor variables.
`degf`	Optional user-specified degrees of freedom for the design. Defaults to one less than the number of replicates.
`...`	Other arguments to `dbConnect`, such as user and password
`con`	object of class `sqlrepsurvey`

For the American Community Survey, scale is 4/80 and rscales is rep(1,80).

The check.factors operation can be slow (eg over an hour for an American Community Survey dataset with 9 million records and 300 variables). If the survey object is saved with save(), it can be reconnected to the database with open, so that it only needs to be created once. The most flexible and fastest approach is usually to create the zero-row data frame manually from the data documentation: only the columns for factor variables need to be supplied, as the code will assume other variables are numeric.

close closes the database connection, first attempting to garbage-collect any tables corresponding to non-existent R objects.

open re-opens the database connection.

an object of class sqlrepsurvey

sqlsurvey, MonetDB

## Not run: 
## assumes data already in database
library(sqlsurvey)
monetdriver<-MonetDB(classPath="/usr/local/monetdb/share/monetdb/lib/monetdb-jdbc-2.4.jar")

alabama<-sqlrepsurvey("pwgtp",paste("pwgtp",1:80,sep=""),key="idkey",scale=4/80,rscales=rep(1,80),
mse=TRUE,database="jdbc:monetdb://localhost/ACS",
driver=monetdriver,user="monetdb",password="monetdb",table.name="alabama3yr",check.factors=TRUE)

## verify against Census Bureau totals
svytotal(~sex,alabama)
svytotal(~I(agep %in% 0:4)+I(agep %in% 5:9)+I(agep %in% 10:14)+I(agep %in% 15:19),alabama)
svytotal(~I(agep %in% 20:24)+I(agep %in% 25:34)+I(agep %in% 35:44)+I(agep %in% 45:54),alabama)
svytotal(~I(agep %in% 55:59)+I(agep %in% 60:64)+I(agep %in% 65:74)+I(agep %in% 75:84)+I(agep>84),alabama)

## other analyses
svymean(~wagp, subset(alabama, !is.na(wagp)), byvar=~sex,se=TRUE)
svyquantile(~agep, alabama,quantiles=0.5,se=TRUE)

plot(svysmooth(wagp~wkhp,alabama,sample.bandwidth=5000))


## with regular expression
alabama<-sqlrepsurvey("pwgtp",repweights="pwgtp[1-9]",key="idkey",scale=4/80,rscales=rep(1,80),
mse=TRUE,database="jdbc:monetdb://localhost/ACS",
driver=monetdriver,user="monetdb",password="monetdb",table.name="alabama3yr",check.factors=TRUE)


## End(Not run)