How computers are useful in data generation?

Answer 1

This question is similar to asking what has the sun to do with light? The essentials behind data generations is the ability to have source data in various forms and to be able to parse it into meaningful structures that represent or codify a certain aspect of logical tree of information in a humanly readable form.

Now the purpose of this data generations can be for hundreds of purposes, (i.e.: formatted output for catalogs, test cases, even for logical pairs matching and others). However, the critical element is to create data in a consistent form that has meaning. Rather than go through all of the scenarios I will will deal exclusively with with just Test Data generation.

In an enterprise test environment where you have external parties providing services or products and are a part of a test environment you will want to test with "production-like" data. The reason you do not use production data itself is that besides the privacy issues if you have to test with segments using production systems you don not want to pollute production data or skew BI assessments.

So using the standard reference data of the company you would generate false data that would mirror the production environment. Some of this is as simple a s scrambling identifiers or replacing leading characters with a known valid character which is excluded from the current data set. The same is in relation personal information. So names can be scrambled and then used as test data with either an indicator in the name or account IDs to isolate it from real production data. This can all be done in moments by a computer rather than days by a human following a simple algorithm.

The second and more complex aspect is the modeling of invalid data that yet looks like real data. While lying is a human trait keeping them consistent and straight is not. Hence the rules for fabrication of false data need to be again developed in algorithms that will test the exception handling of a computer. Some of these errors have to look like keying errors while others need to appear deliberate. As the rules for validating a test are created then a subset of production based modeled data would be transformed to stress the exception handling rules.

Now the reason it has to be human readable is simply that the generated data needs to be in a form that can represent a human input via keyboard or voice that then is interpreted by the testing environments. A computer may actually supply this data.

When taking large volumes of production data, and modeling millions upon millions of records, there is no way to do this humanly in any reasonable time-frame, a computer is necessary.

I hope this answers your question.

Answer 2

Computers are useful in data generation as they can do it much faster than a human could. Jobs, such as accounting, could soon be done mostly by computers.