How do you measure defect density?

Depending up on the specifications,and the how it effects the end user ,and the business of the company.

Defect density is a measure of the total known defects divided by the size of the software entity being measured.

Defect density is defined as the total number of defects identified divided by the total line of code or size. The Number of Defects identified is the count of total defects identified against a particular software entity, during a particular time period.

Some of the Examples are:
- defect to date since the creation of an application
- defects found in a program during an inspection
- defects to date since the release of the software

Defect Density is used to compare the relative number of defects in various software components.

The measure defect-density (dd) is besides the measure loc one of the most widely used measures in industry. The measure dd is defined as: dd = defects (k)loc where defects are the number defects some time after release (normally six months). We will not discuss here the problem of the application of this measure in practice. The next figure illustrates the effects which can occur when using the measure dd. Version 1 version 2 # defects loc defects/loc defects loc defects/loc relation 1 12 777 0.01544 3 55 0.05455 < 2 5 110 0.04545 6 110 0.05455 < 3 2 110 0.01818 3 110 0.02727 < 4 3 110 0.02727 4 110 0.03636 < 5 6 1000 0.00600 70 10000 0.00700 < sum 28 2107 0.01329 86 10385 0.00828 > table 1: behavior of the measure defect-density. Relation shows the relation of the defect-density of version 1 to version 2. Assume, we have two program versions 1 and 2, and each of them consisting of five modules. For each version we consider the measure dd. We see, that every module of version 1 has a lower defect-density than every module of version 2. Summing up the defects and loc of each version then we get the result in the line sum. The result is surprising. Although the value of dd for each of the modules of version 1 is less than the values of dd for version 2 (relation), the value of dd of all modules of version 1 is greater than dd of version 2.