Bayesian Rule Set: A Quantitative Alternative to Qualitative Comparative Analysis

Chiu, Albert; Xu, Yiqing

doi:10.1086/720791

Cited by 2 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, more hyper-parameters will need to be fine-tuned for these algorithms and a valid inference remains a challenging statistical problem. Finally, tree-based methods by construction are more efficient and effective than linear regression in capturing interactions (see detailed illustrations in Chiu & Xu, 2022). In addition to showing efficiency and stability (Speiser et al, 2019), ensemble trees are known for solid performance in large and noisy settings (Sanchez-Pinto et al, 2018) where Lasso underperforms (Hastie et al, 2020).…”

Section: A Comparison With Pattern Discovery By Linear Regressionmentioning

confidence: 99%

Supervised machine learning for theory building and testing: Opportunities in operations management

Chou

Chuang

Chou

et al. 2023

J of Ops Management

View full text Add to dashboard Cite

Machine learning's (ML's) unique power to approximate functions and identify non-obvious regularities in data have attracted considerable attention from researchers in natural and social sciences. The emergence of predictive modeling applications in OM studies notwithstanding, it remains unclear how OM scholars can effectively leverage supervised ML for theory building and theory testing, the primary goals of scientific research. We attempt to fill this gap by conducting a literature review of recent developments in supervised ML in OM to identify vacancies in the extant literature, shedding light on how ML applications can move beyond problem-solving into theory building, and formulating a procedure to help OM scholars leverage ML for exploratory theory development. Our procedure employs the random forest with well-developed properties and inference toolkits that are crucial for empirical research. We then expand the boundary of ML usage and connect supervised ML to the explanatory modeling and hypothesis testing employed by OM empiricists for decades, and discuss the use of supervised ML for causal inference from observational data. We posit that contemporary ML can facilitate pattern exploration and enhance the validity of theory testing. We conclude by discussing directions for future empirical OM studies that aim to leverage ML.

show abstract

Section: A Comparison With Pattern Discovery By Linear Regressionmentioning

confidence: 99%

Supervised machine learning for theory building and testing: Opportunities in operations management

Chou

Chuang

Chou

et al. 2023

J of Ops Management

View full text Add to dashboard Cite

show abstract

“…Finally, in methodological approaches, as illustrated by Chou et al (2023), the utilisation of ML techniques stands out in this study. These techniques consider interactive and complex variables, providing a more efficient and effective means to capture interactions between explanatory variables compared to linear regression, as highlighted by Chiu and Xu (2023). In this context, we endorse ML as a valuable methodology and address recent calls, such as Lindner et al (2022), for additional statistical tools to examine consolidated assumptions, as emphasised by Kalnins (2022).…”

Section: Testing Service Infusion In Manufacturingmentioning

confidence: 88%

“…networks based on the perceptron convergence theorem, constructing networks of neurons and determining class fitting, functions and associated weights aligned with training data (Chiu and Xu, 2023;Hastie et al, 2009;Witten and Frank, 2005).…”

Section: Testing Service Infusion In Manufacturingmentioning

confidence: 99%

“…, 2012). These methods encompass: decision trees , using a top-down divide-and-conquer approach that selects attributes and partitions datasets to create subsets reflecting homogeneity, assessed via entropy (C4.5 algorithm) or Gini criterion (CART algorithm); decision rules , offering rules based on a single attribute (1R algorithm), employing top-down separate-and-conquer approaches (e.g., RIPPER algorithm) or a blend of divide-and-conquer and separate-and-conquer strategies (e.g., PART algorithm); ensemble methods that utilise various learners (e.g., random forest, employing base learner and bagging and boosting through decision trees) weighted to derive a final prediction; nearest neighbours , classifying attributes based on Euclidean distance (e.g., lazy IBk algorithm); Support Vector Machines , crafting hyperplanes to maximise margins between classes and employing kernel functions to establish decision boundaries; statistical classifiers such as logistic regression for probability estimation, Naïve-Bayes estimating class-conditional probabilities by assuming attribute independence, or Bayesian networks specifying attribute independence; and neural networks based on the perceptron convergence theorem, constructing networks of neurons and determining class fitting, functions and associated weights aligned with training data (Chiu and Xu, 2023; Hastie et al. , 2009; Witten and Frank, 2005).…”

Section: Background Literaturementioning

confidence: 99%

See 1 more Smart Citation

Testing service infusion in manufacturing through machine learning techniques: looking back and forward

Bustinza,

Vendrell-Herrero,

Davies

et al. 2024

IJOPM

View full text Add to dashboard Cite

PurposeResponding to calls for deeper analysis of the conceptual foundations of service infusion in manufacturing, this paper examines the underlying assumptions that: (i) manufacturing firms incorporating services follow a pathway, moving from pure-product to pure-service offerings, and (ii) profits increase linearly with this process. We propose that these assumptions are inconsistent with the premises of behavioural and learning theories.Design/methodology/approachMachine learning algorithms are applied to test whether a successive process, from a basic to a more advanced offering, creates optimal performance. The data were gathered through two surveys administered to USA manufacturing firms in 2021 and 2023. The first included a training sample comprising 225 firms, whilst the second encompassed a testing sample of 105 firms.FindingsAnalysis shows that following the base-intermediate-advanced services pathway is not the best predictor of optimal performance. Developing advanced services and then later adding less complex offerings supports better performance.Practical implicationsManufacturing firms follow heterogeneous pathways in their service development journey. Non-servitised firms need to carefully consider their contextual conditions when selecting their initial service offering. Starting with a single service offering appears to be a superior strategy over providing multiple services.Originality/valueThe machine learning approach is novel to the field and captures the key conditions for manufacturers to successfully servitise. Insight is derived from the adoption and implementation year datasets for 17 types of services described in previous qualitative studies. The methods proposed can be extended to assess other process-based models in related management fields (e.g., sand cone).

show abstract

Bayesian Rule Set: A Quantitative Alternative to Qualitative Comparative Analysis

Cited by 2 publications

References 21 publications

Supervised machine learning for theory building and testing: Opportunities in operations management

Supervised machine learning for theory building and testing: Opportunities in operations management

Testing service infusion in manufacturing through machine learning techniques: looking back and forward

Contact Info

Product

Resources

About