GetStratifiedSample: Gets a stratified sample of data from Causata
In Causata: Analysis utilities for binary classification and Causata users.

Description Usage Arguments Details Value Author(s) See Also Examples

Extracts a stratified sample of data

1
2
3

GetStratifiedSample(connect, query, stratification.variable, 
                    stratification.variable.name, 
                    stratification.value=0)

`connect`	Causata connect object - used to resample at the stratified sampling rates.
`query`	Causata query object - used to resample at the stratified sampling rates. Note that the `Limit` must be defined.
`stratification.variable`	A vector of values on which to base the stratification.
`stratification.variable.name`	The name of the Causata variable that is used as the basis of stratification.
`stratification.value`	Value of the stratification.variable which will determine the stratum for a record.

This function gets a stratified sample of data from Causata. The population will be split into two strata based on whether the stratification.variable value for a record matches the stratification.value. Sampling rates for the two strata are then calculated where the rate for the larger strata, strata.A is:

sample.rate.A = sqrt((# records in strata.B) / (# records in strata.A))

New queries are run to resample the Causata data at these sample rates.

Returns a list with two elements as follows:

`df`	A dataframe of sampled data containing all of the variables found in `query`.
`weights`	A vector of weight values. The weights are the inverse of the probability of selecting a record in the sample.

Suzanne Weller <support@causata.com>

Connect, Query, Limit.

# create some variables to query for
variables <- c('customer-id', 'total-spend')

# create a stratified sample given an initial query
# The commands below are commented out since they require an actual server connection
#connection <- Connect(hostname="server.causata.com",
#  username="user@gmail.com", password="enw8Q!mN")
#query <- Query() + Limit(500)
#df <- GetData(connection, query)

# The commands below are commented out since they require an actual server connection
#sampled.data <- GetStratifiedSample(connection, query, 
#  df[['has.purchase__Next.30.Days']], 'has.purchase__Next.30.Days', "true")
#table(sampled.data$weights)