top of page

Algorithms and Their Role In Data Science

Author: Jiahao Wu

Editors: YanXi Chen and Jasleen Matharu

Artist(s): Gianluca Zhang

Algorithms are often mentioned in movies and depicted as complex and high-tech systems beyond our understanding. However, the truth is that they are present in our day-to-day lives and don’t just exist in the sci-fi world. An algorithm is more accurately defined as a set of instructions given to a computer to allow it to do anything essentially: from a simple calculator to complex calculations for firing a rocket.

Algorithms are essential in data science and, more specifically, machine learning. Machine-learning is a type of artificial intelligence that makes software applications more accurate in predicting outcomes without explicitly programming them. Essentially, it “learns” from past data to better infer what the future data will look like. As a result, machine learning helps companies immensely in various ways. For example, it can help predict trends in customer behavior for a specific product or predict equipment failure, which aids the company in becoming more efficient.

In general, there are three ways to characterize machine learning algorithms: supervised, unsupervised, and reinforcement learning. Although these are broad categories, they describe the different ways machine learning can work.

Unsupervised learning is what it sounds like: the algorithm doesn't need to be supervised to work. It discovers patterns and information that were not previously noticeable without human intervention. It forms clusters of different groups through a trait or property without a target variable. Because of this property, unsupervised learning is often used to analyze raw or unlabeled data. Examples of unsupervised learning include Apriori algorithms and K-means.

Supervised learning is similar to unsupervised learning, except the algorithm has a goal. A set of independent variables predicts a specific type of outcome. This algorithm continues until it reaches a sufficient level of accuracy. Rather than unlabeled data, supervised learning uses labeled data to classify/predict outcomes. Examples of supervised learning include regressions.

Finally, reinforcement learning is a method that “rewards” desired behaviors and “punishes” undesired ones. Using a trial and error method, the machine would learn to adapt to its environment and eventually be able to tell the difference between “right” and “wrong.” Examples of reinforcement learning include the Markov decision process and Q learning.

These algorithms are the basics of machine learning and are applied frequently in the artificial intelligence industry (AI). However, these algorithms don’t just appear in codes for AI; they are also instrumental in data science. These algorithms also allow computers to analyze data by themselves, automating the process. Additionally, the various algorithms apply to different data structures and models, which is also included in data science. As a result, algorithms are an essential utility tool for all: used professionally to analyze data and make our lives more convenient through everyday products.



“Common Machine Learning Algorithms.” Analytics Vidhya, Analytics Vidhya, 9 Sep 2017,


“Machine Learning in Data Science” Hevo, Hevo, 29 July 2021,



“Machine Learning in Data Science 2.” Hevo, Hevo, 29 July 2021,



17 views0 comments


bottom of page