In tkonopka/consonance: Consonance Testing

suppressMessages(library(Rcssplot))
library(checkmate)

Introduction

In some situations, it is relevant to perform validations multiple times.

For example, in interactive sessions in which objects can become over-written - by design or by mistake - we might want to check them in the console. Another use case arises in scripts that produce several objects with overlapping requirements.

Individual validations

In order to prep for the next section on validation in batch, let's first review consonance validation on individual objects. Let's set up a consonance suite for data frames.

library(consonance)
suite_numeric <-
  consonance_assert("x numeric", assert_numeric, .var="x", any.missing=FALSE) +
  consonance_assert("y numeric", assert_numeric, .var="y", any.missing=FALSE)

For data objects, we will use two small data frames.

d1 <- data.frame(x=c(3, 4), y=c(7, 0), z=0)
d1
d2 <- data.frame(x=c(3, NA), y=c("7", "0"), stringsAsFactors=FALSE)
d2

The direct way of using consonance is to validate each object against the suite, one at a time.

# should execute quietly
validate(d1, suite_numeric)
# should signal errors
validate(d2, suite_numeric)

Validations in batch

To perform validations in batch, we first specify pairings of objects and test-suites.

batch <- data.frame(object=c("d1", "d2"), suite="suite_numeric",
                    stringsAsFactors=FALSE)
batch

Then, we can validate using this group object

validate(batch)

The batch can be of any size, and becomes increasingly useful when working in environments with a large number of objects. To extend the above example, we can prepare several other data frames.

# additional objects with numeric data
d3 <- data.frame(x=1:3, y=1:3)
d4 <- data.frame(x=1:4, y=1:4)
d5 <- data.frame(x=1:5, y=1:5)
batchB <- data.frame(object=paste0("d", c(1,3,4,5)),
                     suite="suite_numeric",
                     stringsAsFactors=FALSE)
# should execute quietly to signal consonant data
validate(batchB)
````

Quiet execution signals that all data frames pass all the criteria.


## Appendix

```r
sessionInfo()