Approximate questions

HackerEarth has introduced Approximate questions that are similar to programming questions only but the solutions to these problems are more in terms of approximation than a clean precise solution. These types of questions can be used for domains where there is no precise answer as such. For example, image processing or deep vision.


You need to create a custom checker to score approximate questions because these questions have more than one acceptable answer. For example, you need to determine the list of 4 numbers divisible by 2 but less than 100. There are 50 numbers divisible by 2 but less than 100. A candidate can list any combination of 4 even numbers from the list of 50 even numbers and all of them will be correct answers. In the custom checker file, you can choose to give a partial score to a candidate who lists 3 even and 1 odd number. In this case, the candidate will receive 30 points for the question if the total score for this question is 50 points. Therefore, partial scoring can be used in such states where there are one or more answers correct and it depends on the candidates which answer they want to give. It is better than the pass or fail strategy which is used in coding questions where if a test case is passed, you get the full score but if a test case fails, you get nothing.