Inputs to programs generally follow a known distribution. For instance, a health program can expect height to be distributed around the mean height. This tool lets us test the program on its most common inputs and determine which lines of code are run the most.
See webpage for demo.