Learning dispatching rules using random forest in flexible job shop scheduling problems

Jun, Sungbum; Lee, Seokcheon; Chun, Hyonho

doi:10.1080/00207543.2019.1581954

Cited by 65 publications

(11 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Researchers could, for example, fit panel data regression or structural simulation models based on those input features for the sake of more transparency, or use more obscure black boxes, such as deep neural nets, while retaining the benefit of features crafted from theoretical knowledge. For instance, RF models have been shown to effectively improve the performance of job dispatching and outperform widely used dispatching rules for difficult flexible job shop scheduling problems (Jun, Lee, & Chun, 2019). Machine learning performance being a function of the input features engineered by researchers possessing theoretical knowledge of scheduling, we posit that it is contextual knowledge and theoretical understanding that gives OM researchers an edge over computer scientists and statisticians in creating input features and better applying machine learning in operational problems.…”

Section: Generalizabilitymentioning

confidence: 99%

Cross‐item learning for volatile demand forecasting: An intervention with predictive analytics

Chuang

Chou

Oliva

2021

J of Ops Management

View full text Add to dashboard Cite

Despite its importance to OM, demand forecasting has been perceived as a "problem-solving" exercise; most empirical work in the field has focused on explanatory models but neglected prediction problems that are part of empirical science. The present study, involving one of the leading electronics distributors in the world, aims to improve prediction accuracy under high demand volatility for procurement managers to make better inventory decisions. In response to requests for an integrated forecasting methodology, we undertook an iterative process based on three guiding principles-data pooling, theoryinformed feature engineering, and ensemble-based machine learning. The resulting framework managed to improve forecast accuracy significantly and is applicable to a broad range of situations. We present reflections and insights derived abductively through engagement with managers in this problem situation. This "problem-driven" process corresponds to intervention as a research strategy that can foster theoretical and methodological innovations in OM. Our contribution goes beyond the development of the prediction framework as it elucidates ways OM researchers could leverage theoretical foundations to inform feature derivation and model construction. We posit that this work points to a way forward to the combination of OM principles with the emerging innovations in data science and artificial intelligence.

show abstract

Section: Generalizabilitymentioning

confidence: 99%

Cross‐item learning for volatile demand forecasting: An intervention with predictive analytics

Chuang

Chou

Oliva

2021

J of Ops Management

View full text Add to dashboard Cite

show abstract

“…Çömert et al [21] used the random forest model to predict the prediction ability of RF on large data sets. It is found that the prediction ability of the random forest model based on bagging algorithm is better; Jun [22] et al studied the prediction ability of SVM, logistic model, and random forest model. e conclusion is that the prediction ability of random forest model is always better than that of SVM model, while the prediction ability of SVM model with parameter selector is better than that of logistic model [23].…”

Section: Related Workmentioning

confidence: 99%

Empirical Analysis of Financial Statement Fraud of Listed Companies Based on Logistic Regression and Random Forest Algorithm

Liu

2021

Journal of Mathematics

View full text Add to dashboard Cite

Financial supervision plays an important role in the construction of market economy, but financial data has the characteristics of being nonstationary and nonlinear and low signal-to-noise ratio, so an effective financial detection method is needed. In this paper, two machine learning algorithms, decision tree and random forest, are used to detect the company's financial data. Firstly, based on the financial data of 100 sample listed companies, this paper makes an empirical study on the fraud of financial statements of listed companies by using machine learning technology. Through the empirical analysis of logistic regression, gradient lifting decision tree, and random forest model, the preliminary results are obtained, and then the random forest model is used for secondary judgment. This paper constructs an efficient, accurate, and simple comprehensive application model of machine learning. The empirical results show that the comprehensive application model constructed in this paper has an accuracy of 96.58% in judging the abnormal financial data of listed companies. The paper puts forward an accurate and practical method for capital market participants to identify the fraud of financial statements of listed companies and has certain practical significance for investors and securities research institutions to deal with the fraud of financial statements.

show abstract

“…Similarly, Olafsson and Li [27] applied a decision tree to learn directly from scheduling data where the selected instances are identified as preferred training data. Jun et al [28] proposed a random-forest-based approach to extract dispatching rules from the best schedules. This approach includes schedule generation, rule learning, and discretization such that they could minimize the average total weighted tardiness for job shop scheduling.…”

Section: Literature Reviewmentioning

confidence: 99%

A Hybrid Machine Learning and Population Knowledge Mining Method to Minimize Makespan and Total Tardiness of Multi-Variety Products

Qiu

Zhang

2019

Applied Sciences

View full text Add to dashboard Cite

Nowadays, the production model of many enterprises is multi-variety customized production, and the makespan and total tardiness are the main metrics for enterprises to make production plans. This requires us to develop a more effective production plan promptly with limited resources. Previous research focuses on dispatching rules and algorithms, but the application of the knowledge mining method for multi-variety products is limited. In this paper, a hybrid machine learning and population knowledge mining method to minimize makespan and total tardiness for multi-variety products is proposed. First, through offline machine learning and data mining, attributes of operations are selected to mine the initial population knowledge. Second, an addition–deletion sorting method (ADSM) is proposed to reprioritize operations and then form the rule-based initial population. Finally, the nondominated sorting genetic algorithm II (NSGA-II) hybrid with simulated annealing is used to obtain the Pareto solutions. To evaluate the effectiveness of the proposed method, three other types of initial populations were considered under different iterations and population sizes. The experimental results demonstrate that the new approach has a good performance in solving the multi-variety production planning problems, whether it is the function value or the performance metric of the acquired Pareto solutions.

show abstract

Learning dispatching rules using random forest in flexible job shop scheduling problems

Cited by 65 publications

References 25 publications

Cross‐item learning for volatile demand forecasting: An intervention with predictive analytics

Cross‐item learning for volatile demand forecasting: An intervention with predictive analytics

Empirical Analysis of Financial Statement Fraud of Listed Companies Based on Logistic Regression and Random Forest Algorithm

A Hybrid Machine Learning and Population Knowledge Mining Method to Minimize Makespan and Total Tardiness of Multi-Variety Products

Contact Info

Product

Resources

About