Algorithms are everywhere in our daily lives. They find the shortest route to get you to your destination, recommend the next series to watch on your streaming service, and suggest the best move in a game of chess. When we’re dealing with enormous amounts of data and multiple possible choices, algorithms often make better decisions than humans. But algorithms have more advantages which can make them useful in many situations, and not everyone knows about them. Below, I will highlight six of these unknown benefits of using models and algorithms in your decision-making process.
Use real-time information
Computers are much faster calculators than humans. So if you have a problem that needs to be solved before a given deadline, using an algorithm gives you more time before you need to start computing the solution. That means you can work with more up-to-date information, and that in turn will let you make better decisions. Say, you work at a transportation company and it is your job to plan the vans’ routes. If you need to do that by hand, then you would probably have to start a day beforehand. With an algorithm, you could do that on the day itself. Consequently, you have much more information on, for instance, road congestion. Thus, the algorithm improves your planning.
Deal with uncertainty
Although you could always use more real-time information, some data will inevitably remain unknown when making a decision. The time gained by using an algorithm allows you to consider many different scenarios for the uncertain data. Doing that enables you to find the solution that performs the best on average, or makes sure that a certain property is satisfied in 99 percent of possible situations. For example, you operate an electricity network, and the demand of customers is unknown. By considering multiple demand scenarios, you can design a network that satisfies all demands 99 percent of the time.
Perform constantly
A computer does not have a Monday morning blues or a hangover on Friday morning. Moreover, it never needs a holiday. Its performance remains constant throughout the week and the year. This is especially useful if your decisions also influence the operations of your customers. If your schedules always look more or less the same, a customer or supplier will be happier to work with you than with one of your competitors who cannot guarantee consistent quality.
Know your problem
When using models, you need a precise understanding of the problem you need to solve. This might sound like a disadvantage at first, since it can be time-consuming to come up with all the requirements and the objective. However, the exactness of the formulation also prevents misunderstandings between your colleagues. Maybe colleague A thinks that a certain customer always needs to be served before 9.00, while colleague B thinks that 10.00 is also fine. Or maybe colleague A thinks that minimizing the total number of vehicles that are used is the most important objective, while colleague B tries to minimize the total distance that is travelled by the vehicles. If the parameters of the problem are well understood, such differences are addressed from the beginning.
Comply with requirements
Algorithms can integrate hard constraints into your model that always need to be satisfied. So if you use an algorithm to solve your problem, you don’t have to worry about breaking legal requirements or internal policies. If you have the right information in the model, the solutions will always be compliant with those requirements. Furthermore, a model could show whether a solution that you created manually is feasible or not.
Most loyal employee
If you rely on the expertise of humans in your decision-making process, then you are always at risk of losing that knowledge. An employee might get sick or quit their job. An algorithm is the most loyal employee you can imagine and will always be part of your organization. So making algorithms part of your decision-making process is a wise thing to do, especially in a tight labour market.
Hopefully, this list made you think differently about using algorithms in your decision-making process. The advantages are obviously different for each company, so we are happy to discuss the benefits for your specific challenges. If you’ve experienced any of the advantages in the list above, then I would also love to hear that. If you wish to contact me, do not hesitate to send an email to bernard@doingthemath.nl.