Bayesian statistics is in the news again as it has been used to help track the last known locations of the plane MH370. In this particular instance it appears a complex Bayesian analysis was undertaken using a large amount of data to work out the probability of the plane being in any one location as defined by a set of grids across the sea around Malaysia.
Bayesian statistics is good for modelling uncertainty and including some weight to less likely scenarios. It can be used in forensics, healthcare, business modelling or marketing.
Within digital marketing a straightforward application of Bayes’ law gives the chance that an AdWords visitor will make a purchase. This is a simple example and in this case the data is likely to be available via AdWords conversion data as well. But it can also be applied to other sources of sales and give useful data that is hard to obtain by other methods.
The team involved with the Malaysian plane search used this type of mathematical modelling to help with a physical search and bring back the locations with highest probability of finding the missing plane. It finds the best fit from a large amount of sometimes conflicting data.
Another use of Bayesian statistics is within the fields of ‘digital search’ and it appears that this forms part of Google’s search engine algorithm. It returns results with the highest probability of containing the information searched for, and can cope with large amounts of sometimes conflicting data. Understanding Bayesian statistics gives an alternative way of looking at Google search which is helpful for marketing work.
More information on using Bayesian statistics within marketing is on an earlier article called Bayesian statistics - for correct predictions on benefits from AdWords users