Analytics and the art of modeling

Abstract

This paper could have been subtitled “Was Johannes Kepler a precursor of analytics?” Models constitute a key tool for achieving a deep understanding of how complex systems behave, and they have been central to building the body of knowledge that we understand as “science.” Since its origins, operations research has claimed to be a scientific discipline and, as such, it is rooted in the model‐building process. Epistemologically, it is a process that is schematically represented by the methodological chain: facts→laws→theories. One well‐known example of the paradigm of modern science is Kepler's quest to interpret Brahe's astronomical observations in terms of a model, which resulted in his laws of planetary motion. Since then, we have learned that reality is more complex than we thought, and that it is uncertain and dynamic in both probabilistic and chaotic terms. Technological evolution currently provides us with quality data in unprecedented amounts, and this has forced us to develop appropriate methods for dealing with them. However, this only means that the methodological chain for discovering knowledge is now richer with respect to our capabilities of analyzing the facts, and therefore the door is now open to greater possibilities of finding laws and formulating explanatory theories. The search for Higgs boson, frequently presented as an archetype of analytics process, is essentially supported by the same epistemological principles as Kepler's work. This paper is devoted to making some remarks on these similarities and, in doing so, will highlight the relationships between the model‐building processes and the set of techniques under the umbrella of analytics. The above‐mentioned comments took place within the framework of the theme of the IFORS 2014 conference: The Art of Modeling. Relationships are illustrated in a typical, contemporary case, the time‐dependent analysis of mobility patterns in urban areas exploiting information and communications technologies, in which pervasive sensoring makes massive amounts of data available.

Multiple criteria analysis of the popularity and growth of research and practice of visual analytics, and a forecast of the future trajectory

“…Notwithstanding the complexity of today's available big data, decision‐makers can still extract relevant information from the floodgate of the unorganized data through VA and enhance data and systems modeling (Akpan and Brooks, 2012; Barcelo, 2015; Greasley and Edwards, 2019). VA, therefore, combines the automatic processing capability of the computer power with statistical analysis of big data, enhances information visualization and human interaction, which helps to uncover hidden knowledge for problem‐solving and decision‐making (Diakopoulos et al., 2011; Akpan and Brooks, 2012; Chen et al., 2012; de Freitas et al., 2021).…”

Concepts, Applications and Emerging Opportunities in Industrial Engineering

“…Statistics indicate that fatalities and incidents rates in the construction industry remain significantly higher than the all-industry average in many countries. This makes the construction industry one of the most hazardous and accident-prone industries and the majority of the accidents experienced occur due to the inability of the workforce to predict, identify, and respond to hazards at the workplace [38]. The dynamic nature of construction work and task unpredictability on projects make hazard recognition difficult [39].…”

“…Hence, the last decade presented a competitive global market creating a changing work environment that demands engineers who possess soft skills [36]. Further, the current risk management practices require investing more time and effort on the soft skills in order to advance the risk management process [10,22,24,28,37,38].…”

Enhancing discrete-event simulation with big data analytics: A review

“…In general, both analytics and OR are concerned with the collection and analysis of data in order to find patterns for possible explanations or for testing a hypothesis. The distinction lies in that in analytics the role of models is often subsidiary whilst in OR models play an essential role and all other steps are oriented toward assisting the modelbuilding process as well as to test the validity of the model (Barcelo, 2015). Thus whilst analytics is data-driven and has a focus on data and outputs and may have little knowledge of underlying processes, simulation is model-driven with a deep knowledge about processes.…”