Analytical Problem Solving Based on Causal, Correlational and Deductive Models

Abstract

The paper discusses three frameworks for analytical problem solving: causal modelling, correlational modelling and deductive modelling. Table 3 in the paper summarizes the discussion by placing the three types of modelling in the process of analytical problem solving. In this supplemental material, we explain the structure of Table 3 and relate it to the literature.Section 5.1 of the paper discusses essential differences between the three types of modelling. Table 3 organizes these in the problem-solving process. The literatures in various applied fields abound in models of the problem-solving process, typically organized in step-wise procedures. Table 4 collates four such procedures:• Smith (1988), one of the best researched models in the management literature.• Six Sigma's DMAIC model (De Mast and Lokkerbol 2012), widely used and discussed in quality engineering and statistics-driven operations improvement. • The CRISP-DM model in data science (Chapman et al. 2000).• PDCA and Toyota's A3 problem-solving procedure (Marksberry, Bustle and Clevinger 2011), which is representative for many similar problem-solving procedures in practitioner books.

Prediction of Early Periprosthetic Joint Infection After Total Hip Arthroplasty

“…Additional studies are therefore required to establish causal pathways and to identify possible modifiable risk factors. 52 …”

Is designed data collection still relevant in the big data era?

“…When data are observational, it is possible that relationship and patterns are not the direct result of one factor causing changes in the other. This can often cause problems with solutions that assume a causal connection, but when implemented won't necessarily yield the intended results 1 . While there are methods for establishing causal relationships from observational data, 2 they are less direct than a well‐designed experiment and may be subject to misinterpretation.…”

“…This can often cause problems with solutions that assume a causal connection, but when implemented won't necessarily yield the intended results. 1 While there are methods for establishing causal relationships from observational data, 2 they are less direct than a well-designed experiment and may be subject to misinterpretation. With a controlled experiment, if the inputs are carefully manipulated and the responses change, there is much more evidence of a causal connection between the inputs and outputs, which can be more directly leveraged to improve understanding and the performance of the system.…”

Perspectives on Curricular Priorities for Big Data Analytics: A Qualitative Study of Educators and Practitioners

“…Indeed, deriving actionable and novel insights from Big Data requires a multifaceted framework involving descriptive, diagnostic, predictive, and prescriptive analytics (de Mast et al, 2022) and necessitates expertise beyond traditional technological and statistical tools and skills. Specifically, this encompasses "computational, algorithmic, statistical and mathematical techniques" (Leonelli, 2020, para.…”