One of the problem data miners are often confronted is how to present statistics to the general audience. It can be the public, other researchers, decision makers, etc. This issue is discussed in a recent article in Decision Science News entitled “Some ideas on communicating risks to the general public”. The article discusses issues when sharing probabilities, conditional probabilities, risks, frequencies, etc. with the public.
The author provides excellent examples to show how probabilities can be ambiguous. A simple statement such as “The probability that it will rain on January 1st is 30%” is already ambiguous. It can be interpreted in several different ways. In addition, it is not clear what rain means (one drop, one minute of rain, etc.).
The author also explains conditional probabilities with a simple example: cancer statistics. He writes that, given the needed numbers, most doctors fail when asking to estimate the probability that a person from the population who tests positive actually has cancer. I think this is quite scarying that the people reading and interpreting medical results don’t know how to come to these results.
Read the full article and tell us your opinion about it.