numero.clean {Numero}

Clean datasets

Description

Sets row names and removes unusable columns and rows.

Usage

numero.clean(..., identity = NULL, na.freq = 0.9,
             num.only = TRUE, select = "")

Arguments

...

Matrices or a data frames.

identity

Name(s) of the column(s) that contain identification information.

na.freq

The proportion of how many missing values are allowed in each column and in each row.

num.only

If true, only numeric columns are included.

select

Indicate if only identities present in all datasets or in exactly one of the datasets are included.

Details

If multiple identity columns are provided, composite identity keys are constructed by concatenating elements from each column with "_" added as a separator.

The frequency of missing values (against na.freq) is tested first by column then by row.

Selection can take three values: "" for no selection, "union" for all identities expanded to every dataset, "shared" for only those data points present in all usable datasets or "distinct" for excluding any points that can be found in more than one dataset. Note that the union may result in rows with no usable values.

Value

A data frame if only one input dataset, or a list of data frames if multiple datasets.

Author(s)

Ville-Petteri Makinen

Examples

# Import data.
fname <- system.file("extdata", "finndiane.txt", package = "Numero")
dataset <- read.delim(file = fname)

# Create new versions for testing.
dsA <- dataset[1:250, c("INDEX","AGE","MALE","uALB")]
dsB <- dataset[151:300, c("INDEX","AGE","MALE","uALB","CHOL")]
dsC <- dataset[201:500, c("INDEX","AGE","MALE","DIAB_RETINO")]

# Select all rows.
results <- numero.clean(a = dsA, b = dsB, c = dsC, identity = "INDEX")
cat("\n\nNo selection:\n")
print(nrow(results$a))
print(nrow(results$b))
print(nrow(results$c))

# Select all rows and expanded for all identities.
results <- numero.clean(a = dsA, b = dsB, c = dsC, identity = "INDEX",
                        select = "union")
cat("\n\nUnion:\n")
print(nrow(results$a))
print(nrow(results$b))
print(nrow(results$c))

# Select only rows that are shared between all datasets.
results <- numero.clean(a = dsA, b = dsB, c = dsC, identity = "INDEX",
                        select = "intersection")
cat("\n\nIntersection:\n")
print(nrow(results$a))
print(nrow(results$b))
print(nrow(results$c))

# Select only rows with a unique INDEX ('dsB' has none).
results <- numero.clean(a = dsA, b = dsB, c = dsC, identity = "INDEX",
                        select = "exclusion")
cat("\n\nExclusion:\n")
print(nrow(results$a))
print(nrow(results$b))
print(nrow(results$c))

# Add extra identification information.
dsA$GROUP <- "A"
dsB$GROUP <- "B"
dsC$GROUP <- "C"

# Select rows with a unique identifier.
results <- numero.clean(a = dsA, b = dsB, c = dsC,
                        identity = c("GROUP","INDEX"),
                        select = "exclusion")
cat("\n\nMulti-identities:\n")
print(nrow(results$a))
print(nrow(results$b))
print(nrow(results$c))