When Data Analytics Goes Wrong
We read an interesting article on the Internet recently which succinctly describes how data analytics can easily lead to wrong conclusions.
The illustration was simple and eye-opening: A man with only two hours’ sleep the previous night rear-ends a car at a red light. Based on the data thus far, we conclude that his lack of sleep caused a lack of attention.
This is an example of a wrong conclusion drawn from data analysis. The deficiency of the analysis is that the data did not log all possible factors that had a bearing on the mishap. For instance, was the man busy tweeting from his phone at the time? Was there a defect in the road? Did the brakes fail?
The lesson here is that prior to embarking on a data analysis project, we should make sure the data is comprehensive. If it isn’t, either draw conclusions with a rider mentioning which factors were not considered, or declare that the data is insufficient for the task (this might cost you additional revenue, but will preserve your reputation).
