Operations Research and Machine Learning

- Overview

Operations Research (OR) can be called a subset of Artificial Intelligence (AI) (at least to some extent, both use data to support decision-making processes), which naturally includes Machine (and Deep) Learning (ML). However, OR and ML are often viewed as different and often alternative approaches to data-driven decision-making, often considered part of the broader field of data science (DS). 

A very simplistic view might suggest that DS dominates descriptive analytics, ML is the reference method for predictive analytics, and OR is known as the dominant method for prescriptive analytics. Reality is obviously more complicated than this!

We've all seen that DS is becoming increasingly "sexier" among board members in business organizations, with ML and DL becoming buzzwords that attract laypeople and general management, while operations rarely share this realization. 

DS and ML often provide (relatively) fast tools to help users understand the ever-increasing volumes of data that companies have to deal with, which is clearly an advantage over OR, which often starts with the modeling of complex processes.

- Synergies between OR and ML

Operations Research (OR) and Machine Learning (ML) are both used to support decision making processes. OR is used to solve complex problems related to systems, such as businesses and networks of machines. ML can be used to look for patterns in data sets with no pre-existing labels. 

Combining OR and ML can produce benefits, such as: 

• Enhanced accuracy
• Optimal solutions
• More dependable predictive model solutions

OR and ML are often seen as distinct approaches. However, there is a huge synergy between them. For example, some ML researchers use OR to improve their learning.
OR uses mathematical methods to solve a numerical version of a problem. ML can speed up research and enable new types of analyses.

- Operations Research vs. Machine Learning

Operations research (OR) and machine learning (ML) are two research fields that have advanced independently for decades. However, they are closely connected and can be combined to create accurate and optimal solutions. 

Here's some related information about OR and ML:

• Operations research (OR): Focuses on optimization, and is concerned with a large collection of unique methods for specific classes of problems. OR researchers have developed techniques to find the optimal value of decision variables to minimize or maximize an objective function. OR can help when an organization has a problem with huge numbers of possible combinations or lots of randomness. For example, airlines use operations research to decide where to fly and when.
• Machine learning (ML): Is used to build predictive models. ML can provide inputs to OR.
• Combination of OR and ML: The two approaches can be combined in a "predict-then-optimize" paradigm. The ML model predicts probabilities or regression outcomes, which are then fed into an optimization model. The optimization model then makes coordinated decisions.

- Operations Research vs. Data Science 

Operations research (OR) and data science are both fields that use data and advanced analytics to make decisions. 

However, they have different areas of focus:

• OR is a scientific discipline that uses mathematical models to optimize business operations. Data science is an interdisciplinary field  that uses statistics, scientific computing, and algorithms to extract insights from data.
• OR is concerned with optimization. Data science is mainly about finding information via data.
• OR methodology offers the capability of testing multiple scenarios of behavior within complex, highly variable systems. Data science does not offer a causal explanation of any relationships between the variables.

- AI in Operations

Operations research (OR) is part of artificial intelligence (AI) because both use data to support decision-making processes. OR and machine learning (ML) often work together in a "predict-then-optimize" paradigm. 

In this paradigm, a machine learning model predicts probabilities or regression outcomes, which are then fed into an optimization model. The optimization model then makes coordinated decisions.

AI operations and optimization involves the application of AI technologies, such as machine learning and advanced analytics. This is done to:

• Automate problem-solving and processes in network and IT operations

• Enhance network design and optimization capabilities

• Enhance efficiency, productivity, and decision-making processes