- To create algorithms for a given problem several stages must be followed:
- The problem must be sufficiently analysed and understood. The requirements and purpose of the algorithm must be clearly stated
- The problem must be sufficiently decomposed into sub-problems. A complex problem may require many sub-problems with each sub-problem requiring their sub-problems and algorithms. When determining an algorithm for a sub-problem developers need to consider inputs, outputs, processes and long term data storage (if necessary)
- Consideration must be given to how data is going to be gathered, stored and processed as well as how output is going to be displayed to the user
- For example, will data be entered via input boxes or file upload/opening? Is data to be stored in a file? What overall processes will be performed on the data? Will the data be displayed on a form, a web page or output into a file?
- The structure of the problem and subproblems should be illustrated by creating a structure diagram which details the breakdown and hierarchy of sub-problems
- Once each sub-problem is specified, an algorithm can be created using flowcharts or pseudocode. The algorithm should be clear, easy to read and as simple as possible
- Meaningful variable names help readers understand what data is being stored and their purpose
- Mathematical symbols such as “=” or “<=” are quicker and simpler to read than “equal to “ or “less than or equal to”. “Percentage < 0 OR Percentage > 100” is clearer than “Percentage under zero or Percentage over one-hundred”. Be concise
- Sets of predetermined test data should be dry run on the algorithm, including normal, abnormal, extreme and boundary data. The results should be displayed in a trace table, with clear outputs to allow errors to be detected
- Encountered errors should be fixed and the algorithm retested to make sure it functions as intended
