2.2.2

Question 1

Problem Recognition

Answer 1

• Stakeholders say what they need from the solution.
• This information is used to produce a clear list of system requirements and a
  definition of the problem.
• We may consider the strengths and weaknesses of a current system.
• We may consider the required inputs, outputs and the volume of stored data.

Question 2

Problem Decomposition

Answer 2

• The problem is broken down into smaller
  subproblems.
• This is repeated until each subproblem can be represented using a single subroutine.
• This reduces the complexity of the problem and makes it easier to solve.
• It enables programmers to see which areas can be solved using pre-existing libraries or modules.
• It makes the project easier to manage.
• Subproblems can be assigned to different specialist teams or individuals.
• Modules can be designed and tested individually before being combined.
• It makes it possible to develop modules in
  parallel and therefore finish more quickly.
• It is easier to debug the code and locate errors.

Question 3

Divide and Conquer

Answer 3

• A problem solving technique with three parts.
• Divide - halve the size of the problem with each iteration.
• Conquer - solve the subproblems.
• Merge - combine the solutions.
• It is applied in binary search, quick sort and merge sort.
• It is a quick way to simplify complex

Question 4

Abstraction

Answer 4

• Represents real world entities using computational elements.
• Excessive details are removed to simplify
  the problem.
• This may then match a problem which
  has previously been solved.
• Existing modules, functions or libraries
  can then be used to solve the problem.
• Levels of abstraction divide a complex
  problem into smaller parts.
• Different levels can be assigned to teams
  whilst hiding details of other layers.
• This makes the project easier to manage.
• Abstraction by generalisation groups together sections with similar functionality.
• This allows segments to be coded together, saving time.

Question 5

Backtracking

Answer 5

• Uses algorithms, often recursively.
• Builds a solution methodically.
• Based on paths which have been visited and found to be correct.
• The algorithm backtracks to the previous stage if an invalid path is found.

Question 6

Data Mining

Answer 6

• Identifies patterns or outliers in large data sets, often collected from multiple sources.
• These data sets are known as big data.
• It spots correlations between data and other trends which might not be easy to see.
• Can be used to make predictions about the future.
• A useful tool to assist in business and marketing.

Question 7

Heuristics

Answer 7

• A non optimal or rule of thumb approach.
• Used to find an approximate solution to a problem.
• Used where the standard solution takes too long.
• Does not produce a 100% accurate or complete solution.
• Provides an estimate for intractable problems.
• Performance Modelling
• Mathematical method to test loads on systems.
• A cheaper and less time consuming method of testing applications.
• Used for safety critical systems where a trial run can’t be carried out.

Question 8

Pipelining

Answer 8

• Modules are divided into individual tasks.
• Tasks are developed in parallel.
• Allows faster completion.
• The output of one process is often the input of another.
• Often used in RISC processors, which perform different parts of the Fetch-Decode-Execute cycle at the same time.

Question 9

Visualisation

Answer 9

• Presenting data using charts or graphs.
• Makes it easier for humans to understand.
• Allows trends or patterns to be more easily identified.