In a lot of researches, statistics are displayed with numerical values, percentages, probabilities, or even simple numbers.

It is also a good way that shows and explains basic metadata, distribution of categories.

Some specific metrics for a particular domain are applied. For example, in Healthcare there is a set of pre-defined metrics to use for persons' medical results measurement. If we talk more about technologies, Application performance monitoring (APM) has also specific metrics such as CPU usage, throughput (request per minute), and others.

Raw numbers are taken as input and as output, specific measurement value/metrics are got.

The most popular is the percentage and numbers that show quantity. But both of them have some advantages and disadvantages.

What is Your point of view?

Let's suppose we have a system that processes requests.

  • The number of error requests is equal to 10, and we assume that it is a low number and that is good, but in fact, the system has processed only 10 requests and all of them have errors.
  • We observe that the error rate during the last hour (percentage of errors that occurred) was very high: 75%. But as it turned, the were 3 as error and 4 as total requests. So, we have an issue, but it is not very critical and a low number of requests faced it.
  • Consequently, are there any alternatives to show the value in the best way to avoid bias and misleading conclusions?

    More Taras Ustyianovych's questions See All
    Similar questions and discussions