R/Test

From QERM Wiki
(Difference between revisions)
Jump to: navigation, search
m (added category link)
 
Line 1: Line 1:
==RUnit Overview==
+
==Overview==
  
 
Suppose you've written a function in R.  You want to make sure it works, so you try it with some sample data.  By saving that information as a test case, you can re-run it again later if you ever need to change the function (for example, to make it run faster, or fix a case you hadn't thought of before).  [http://runit.sourceforge.net/ RUnit] is a package that makes this easy.  It's an [http://en.wikipedia.org/wiki/XUnit xUnit] style framework (don't worry if you don't know what that means). You can find a nice introduction to RUnit [http://cran.r-project.org/web/packages/RUnit/vignettes/RUnit.pdf here].
 
Suppose you've written a function in R.  You want to make sure it works, so you try it with some sample data.  By saving that information as a test case, you can re-run it again later if you ever need to change the function (for example, to make it run faster, or fix a case you hadn't thought of before).  [http://runit.sourceforge.net/ RUnit] is a package that makes this easy.  It's an [http://en.wikipedia.org/wiki/XUnit xUnit] style framework (don't worry if you don't know what that means). You can find a nice introduction to RUnit [http://cran.r-project.org/web/packages/RUnit/vignettes/RUnit.pdf here].
 +
 +
Another unit-testing package is testthat, which is written by the famous Hadley Wickham. It's new and I (Gregor) haven't tried it yet, but given Hadley's other work (ggplot2, plyr) it's probably pretty darn good.
  
 
==Step-by-Step==
 
==Step-by-Step==

Latest revision as of 20:00, 31 October 2011

Overview

Suppose you've written a function in R. You want to make sure it works, so you try it with some sample data. By saving that information as a test case, you can re-run it again later if you ever need to change the function (for example, to make it run faster, or fix a case you hadn't thought of before). RUnit is a package that makes this easy. It's an xUnit style framework (don't worry if you don't know what that means). You can find a nice introduction to RUnit here.

Another unit-testing package is testthat, which is written by the famous Hadley Wickham. It's new and I (Gregor) haven't tried it yet, but given Hadley's other work (ggplot2, plyr) it's probably pretty darn good.

Step-by-Step

Here's the quick summary of what you need to do:

  1. Install RUnit from CRAN
  2. Create a file for the tests named runit.yourfilename.R (the important thing is that it start with "runit", though that is configurable)
  3. Create test functions that start with "test" (also configurable)
  4. Add tests using the checkEquals() and checkException() functions
  5. Run the tests as follows (you can also save to a file):
library( RUnit )

testsuite = defineTestSuite( "name", dirs="directory where your test files are" )

printTextProtocol( runTestSuite( testsuite ) )

Example

Let's use a function that calculates exponential population growth as an example. Suppose it's saved in a file called model.R. Here's the function:

calculatePopulation = function( initialSize, rate, time )
{
    if ( initialSize < 0 )
        stop( "initial size must be positive")
		
    return( initialSize * exp( rate * time ) )
}

Here's an example runit.model.R file:

test.computePopulation = function()
{
    checkEquals( calculatePopulation( 1000, 0.2, 0 ), 1000, "time 0")
    checkEquals( calculatePopulation( 1000, 0.2, 10 ), 7389.056, "time 10")
    checkEquals( calculatePopulation( 1000, -0.2, 10 ), 135, "negative rate", tolerance = 1 )
    checkException( calculatePopulation( -1000, 0.2, 100 ), "negative initial size" )
}

The checkEquals() function takes the actual output from your function and compares it to the expected result you provide. You can also provide a message, which helps keep track of what you're testing. The checkException() function let's you check for expected errors.

Personal tools