Finding a Good Hotel with Chernoff Faces
I’ve been wondering whether using Chernoff faces might be a good variation of the “advanced search” pattern in the context of finding a hotel to stay in.
Choosing the right hotel requires a number of quite complicated things to be considered. But which things you place the most emphasis on depends very much on the context of why you are booking a hotel in a given location or time. If price is the only consideration you have, then you’re lucky. The hotel star rating; the distance of the hotel to where you want to spend your time; the opinions of other people; photos; the existence or absence of certain amenities (gym, pool, etc.) – all and more of these things usually come in to play to an extent.
Chernoff faces are one way of encoding easily decoded multivariate statistical graphics. Humans are also very good at spotting minute variations in human faces, and pictures of faces possess the obvious quality of being instantly recognisable as such. Hotels posses some data ranges that could be encoded into facial attributes to spot outliers, so I’ve put this all together and done a quick sketch of how this might work in practice.
Firstly, some things to note about the design here. I’m assuming that most people place their primary emphasis on price and one other dimension such as star rating or customer rating. Hence the x/y scales to establish the initial plot of the faces (I’ll call these the “primary dimensions”). For many people, that might be enough without the use of facial attributes to reveal further detail.
One practical issue in the UI here is occlusion. What if several hotels posses the same primary dimensions? In that case, some form of cluster management will have to take place – for example on mouse over, the overlapping faces will spring apart to reveal their individual features.
Another issue is interpretation. It does require a bit of work. But this is after all an “advanced search” technique not aimed at the casual customer. The use of Chernoff faces in this way does, I think, allow you to find the best hotel for you in a way that would be difficult if not impossible to achieve in any other way.