Accuracy Requirements
A
accuracy requirement is a
correctness
requirement that specifies a required amount of accuracy,
which is a
quality factor that is defined as follows:
- Accuracy
- the magnitude of defects (i.e., the deviation of the
actual or average measurements from their true value) in
quantitative data.
The typical objectives of a accuracy requirement are to:
- Specify the maximum permitted magnitude of defects in
numerical data.
- Ensure that numerical data represents the true values of
the data.
Accuracy requirements are typically specified in terms of
the following measurements:
- Confidence Interval.
- Standard Deviation.
- Maximum Error.
The following are typical examples of accuracy
requirements:
- “The value of all monetary amounts shall be
accurate to the nearest smallest denomination for that
currency (e.g., penny for $USD).”
- “The value of the temperature of the chemical
mixing tanks shall be accurate to within 1.5°
Celsius.”
- “All timestamps associated with user interactions
shall be accurate to the nearest second.”
- “All reoccurring batch processes shall begin
execution within 10 minutes of their regularly scheduled
start times.”
The following guidelines have been found to be useful when
producing accuracy requirements:
- The scope of an accuracy requirement can be:
- Accuracy requirements can be identified and specified in
term of the following:
| Component of
Requirement |
Possibile Values |
| Numerical Variable |
TBD |
| Measurement Type |
Confidence Interval
Standard Deviation
Maximum Error |
| Error Tolerance |
Maximum Value |
- Accuracy requirements do not refer to character data. For
example, The minimum size of textual data should be specified
as a data requirement rather than an accuracy
requirement.
- Numerical data can be accurate without being precise, and
precise without being accurate.
- For example, a temperature sensor is accurate to 0.1°
Celsius if the temperature measurement returned by the
temperature sensor is always within 0.1° of the true
temperature.