The term “algorithm” is frequently mentioned when discussing artificial intelligence or data analysis. But what exactly is an algorithm and what are its applications?
An algorithm is a set of instructions that defines how to perform a specific task.1
We can compare an algorithm to a recipe to follow. In fact, the terms “algorithms” and “models” are often used interchangeably in the field of data analysis.
Different Types of Algorithms
Just as there are an infinite number of recipes, there are also a large number of algorithms. These can be very simple, such as determining if a number is even or odd, or more complex, such as predicting the decisions of the Supreme Court justices of the United States.2
Descriptive Algorithm
The first type of algorithms is descriptive. These algorithms describe datasets by identifying patterns or converting images and text into numbers. For example, a descriptive algorithm can be used to categorize movies by genre and then create a personalized movie recommendation engine.
Predictive Algorithm
The second type of algorithms is predictive. These algorithms forecast an outcome, as in the previously mentioned example of the Supreme Court. Another famous example of a predictive algorithm is the Moneyball case, where baseball batting averages are used to predict a player’s contribution to their team.3
Prescriptive Algorithm
Finally, the last type of algorithm is the prescriptive algorithm. These algorithms recommend the best action to take, such as which insurance policy to choose or which investments to make. Prescriptive algorithms are particularly useful in a business context as they are decision-support tools in complex situations.
Example of Business Application
You are a large enterprise and you want to focus your activities and sell some of your divisions. The amount of data to analyze is immense, and the possibilities are numerous. A prescriptive algorithm could take into account your constraints (e.g., if I sell A, I must also sell B because they are complementary, or if I keep C, I need to invest $X to modernize it and my maximum budget is $Y) and provide you with the best recommendation.
How to Choose the Right Algorithm?
As the previous examples demonstrate, algorithms have several useful applications and their complexity can vary greatly, from simple decision trees to artificial intelligence models. Choosing the right algorithm, or model, depends on the problem you are trying to solve and your context.
The first step is to clearly define the problem, as recommended by the Six Sigma method. Afterwards, you can get assistance in choosing the right model to achieve the desired result. It is important to remember that the simpler the model, the easier it will be to explain.
- Dimitris Bertsimas, Applied Business Analytics, MIT, 2022. ↩︎
- Daniel Martin Katz, Using data to predict Supreme Court’s decisions, https://msutoday.msu.edu/news/2014/using-data-to-predict-supreme-courts-decisions, Michigan State University, 4 novembre 2014. ↩︎
- L’IA générative n’a pas été utilisée dans l’idéation, la rédaction ou la révision de cet article. ↩︎
Generative AI was not used in the ideation, writing, or review of this article.
