Much has been written about Santa’s achievements on Christmas Eve in terms of the distance and speed that he travels, the weight of gifts that he delivers, and the number of mince pies that he consumes along the way. But without an impressive array of Artificial Intelligence algorithms underpinning his efforts, none of it would be possible.
The logistical challenges are, of course, formidable. In order for Santa to do his job a customer database is needed, one that contains many billions of addresses and is 100% accurate at exactly the same time every year.
Even though many of Santa’s wish lists are posted on Christmas Eve, next day delivery is guaranteed, with a delivery window of perhaps two hours between midnight and 2 am. And none of the packages are ever left with a neighbour, or just dumped on the door step.
But, not everyone provides Santa with a wish list. Therefore the first, and perhaps the most important of the algorithms is for ‘gift preference prediction’.
For this algorithm to work the customer database needs to include a record of all previous gifts, purchases, likes, dislikes and browsing histories, not to mention a record of behaviour (specifically incidences of being ‘good’ or otherwise).
A suitably named ‘logistic regression’ algorithm will produce the necessary gift preferences, ranked in order of their ‘fit’ with the each of the customer characteristics contained in the database.
All the gifts however need to fit into the sleigh. Therefore, running in parallel with this algorithm is one that adjusts the preference list if the total volumetric weight of all the gifts steps outside set boundaries. The sleigh may be extensive in size but it is assumed to be of finite volume.
A suitable choice is the ‘Knapsack’ algorithm. It figures out the maximum number of objects that can be fitted into a set volume. This algorithm will adjust the choices made by the gift preference algorithm in order that the maximum volumetric weight from all the gifts can be accommodated within the sleigh.
These two algorithms need be run many times over. Last minute wish lists have to be accommodated as do last minute fluctuations in behaviour, the latter being used to rule a potential recipient into, or out of the gift delivery schedule. This means that, once the final list is confirmed and signed off, there may be only a few minutes within which to load many millions of tons of gifts on to the sleigh in exactly the right delivery sequence.
The next algorithm concerns the routing of the sleigh. Given the extraordinarily tight delivery window the priority will be to find the shortest path through all destinations.
Dijkstra’s algorithm is perhaps the most suitable for the task. It is used widely in routing protocols to find the shortest path through, for example, a complex road network.
Lastly, there is the need for a load balancing algorithm similar to that used with aircraft. During the trip weight will be lost from the back of the sleigh as gifts are distributed, but gained at the front as Santa eats several million tons of mince pies.
There are few references to Santa’s sleigh requiring ballast movement, but this will probably be essential for stability.
The IT implications are significant. If limited to a single machine, and given the size and scope of the task, the preference prediction algorithm would require many days to run. The Knapsack and Dijkstra’s algorithms would both require many years. All the algorithms would, however, need to be run many times over to capture last minute dips in behaviour as well as the last minute Santa wish lists.
In order to reduce processing times down to a few seconds, an array of parallel quantum processors would be required that could only be assembled in space. And the data that would be needed, about our choices, behaviours, thoughts and wishes would be vast, dwarfing (or should it be ‘elfing’) anything that any of the world’s intelligence agencies could be capable of collecting.
This data would of course be a gold mine to any intelligence agency or commercial enterprise.
But there hasn’t been a word about any form of cyber attack or leak, so the data seems to be totally secure. And every year the whole enterprise works like clockwork (apparently), not to mention on time.
If only we could all operate like Santa.
Happy Christmas to all our readers.
We have long experience of helping organisations to operate like Santa, and all the year round – as analytics are not just for Christmas. If we can do the same for you, then please give us a ring.