Today we explore the concept of optimisation and how it can, quite literally, make our lives a little better. After all, this is what optimisation is about, finding the “best” outcome possible, whatever “best” means in each situation.
This is the third article in my “Learning from Algorithms” series. If you haven’t done so already, I recommend you read Part 2 on Different Costs before this one as we will be referring back to some of its concepts in this article. You can also find the first article on Input/Output here.
Algorithms are great at optimising. They increase engagement, decrease cart abandonment, or give you the most accurate weather forecast. It is also an algorithm that assigns a driver to deliver your mid week pad thai and decides which price to display for the next flight you want to book. Each algorithm has a different objective to optimise. The food delivery algorithm thrives for the shortest delivery times, the pricing algorithm wants to maximise profit for the airline.
While these examples seem rather clear cut, there are often competing interests at play that could be optimised for instead. While keeping delivery times short is a key objective, it might be useful to combine orders to similar areas to decrease cost, even if delivery times become longer. Airlines may want to sell tickets quickly, and hence cheaper, as it means more planning security for them.
If you have read the second article in this series these competing interests may sound familiar to you. They are the different costs associated with an action. When optimising we need to be clear on what we want to optimise for, i.e. which cost to increase/decrease.
In our daily lives this means to be explicit about what we want to change and why. Let’s come back to the grocery example. Neither my partner nor me particularly like grocery shopping. When it is inevitable we both desire an optimised shopping experience. One solution could be to cut the shopping time as short as possible. This minimises the time spent on an undesirable task as we rush through the aisles as fast as we can.
My approach is a little different. I’d rather use the time in a physical supermarket to explore new products and to discover new brands. This might even take more time than a standard shop, but it increases my overall enjoyment.
When optimising we might also take other costs into account, for example optimising one cost while keeping other costs below a certain point. This is so-called “constraint optimisation”. In the shopping example my partner might agree to the exploring, but only if we keep the overall time spent in the supermarket under 45 minutes.
They key lies again in clear communication and transparency, both with others and yourself. The concept of optimisation can help you articulate your needs and focus such a conversation. The following questions can help you with that:
- What cost do you seek to decrease first and foremost?
- Are there any other costs you want to consider?
- How can you achieve this?
Stay tuned for the next part of this series where we talk about loss functions!