**Rebecca**, we did spend a bit of time and effort figuring out how to best group activities into meaningful groups. We had a good bit of preliminary data to find which activities tended to group together – but trends are just trends, there are always exceptions. A great example is the Tracking category where some of the specific activites are really only good for small dogs and some others may be best for large dogs. But the temperaments or personalities of dogs that benefit most from tracking activities have some similarities despite their body sizes being completely different. So we don’t expect that *all* the activities in a highly rated category will be great for the dog they are recommended for – but it’s likely that one of them could be.
There is always a bit of a balancing act between binning and splitting this type of data. If we bin too much we get activities that really might not be similar at all grouped together. On the flip side, if we split out each individual activity, we’d not be able to make useful inferences: you’d tell us your dog likes Fly Ball, and we’d be able to say “Your dog may like Fly Ball” – not so helpful. But by finding clusters of activities that have grouped together in many dogs (we did a cluster/factor analysis on preliminary data) we can make inferences that many dogs that like Fly Ball also like Agility – and we can also see what temperament or personality traits from the surveys correlate with that sport preference.
**Lindsay**, you’re absolutely right: for the moment the recommendations may be pretty bad. Help us improve them. As noted on the sports page:
> Please note that the scores for these recommendations (the number of “paws”) will change as we gather more data on which to base the recommendations. For now these are quite preliminary
But even with an expansive data set, we’ll still be wrong from time to time. Life would be boring if it could be so easily reduced to a handful of numbers. But that doesn’t mean the numbers aren’t useful. These sports recommendations work a lot like career recommendations based on personality tests for people. We may know that *on average* personality types like an INTJ on the Myers Briggs scale may be well suited to carreers in the scienes. So all else being equal, and INTJ should think whether the sciences may be right for them. But there are countless INTJs who would be miserable as scientists, and many phenomenal scientists who have Myers Briggs types that are not common in the field.
I’ve also written up an article all about these sports recommendations for our November newsletter which unfortunately has been delayed. But that should be out soon.