Searching for rules to detect defective modules: A subgroup discovery approach

Rodríguez, Daniel; Ruíz, Roberto; Riquelme, José C.; Aguilar–Ruiz, Jesús S.

doi:10.1016/j.ins.2011.01.039

Cited by 55 publications

(18 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, these methods focus only on maximizing classification accuracy and aim to achieve interpretability just by building the model from rules. Similarly, algorithms for problems such as subgroup discovery [24, 33, 40, 46], contrast set learning [3, 4, 30], and emerging pattern mining [17, 19] identify sets of rules to describe the relationships among variables and discover interesting patterns in the data. In contrast, our work explicitly defines an objective function that scores interpretability and accuracy, and by optimizing it, we find a globally near-optimal model.…”

Section: Related Workmentioning

confidence: 99%

Interpretable Decision Sets

Lakkaraju

Bach

Leskovec

2016

Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

457

View full text Add to dashboard Cite

One of the most important obstacles to deploying predictive models is the fact that humans do not understand and trust them. Knowing which variables are important in a model’s prediction and how they are combined can be very powerful in helping people understand and trust automatic decision making systems. Here we propose interpretable decision sets, a framework for building predictive models that are highly accurate, yet also highly interpretable. Decision sets are sets of independent if-then rules. Because each rule can be applied independently, decision sets are simple, concise, and easily interpretable. We formalize decision set learning through an objective function that simultaneously optimizes accuracy and interpretability of the rules. In particular, our approach learns short, accurate, and non-overlapping rules that cover the whole feature space and pay attention to small but important classes. Moreover, we prove that our objective is a non-monotone submodular function, which we efficiently optimize to find a near-optimal set of rules. Experiments show that interpretable decision sets are as accurate at classification as state-of-the-art machine learning techniques. They are also three times smaller on average than rule-based models learned by other methods. Finally, results of a user study show that people are able to answer multiple-choice questions about the decision boundaries of interpretable decision sets and write descriptions of classes based on them faster and more accurately than with other rule-based models that were designed for interpretability. Overall, our framework provides a new approach to interpretable machine learning that balances accuracy, interpretability, and computational efficiency.

show abstract

Section: Related Workmentioning

confidence: 99%

Interpretable Decision Sets

Lakkaraju

Bach

Leskovec

2016

Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

457

View full text Add to dashboard Cite

show abstract

“…The KC2 dataset was a data processing project developed with C ++ language and included 523 modules, of which 15% were defective. The KC3 dataset was a storage management project developed with Java language and included 458 modules, of which 9% were defective [20]. The CM1 dataset was a NASA spacecraft instrument project developed with C language and included 498 modules, of which 10% were defective.…”

Section: System Descriptionmentioning

confidence: 99%

The use of cross-company fault data for the software fault prediction problem

Çatal¹

2016

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Abstract:We investigated how to use cross-company (CC) data in software fault prediction and in predicting the fault labels of software modules when there are not enough fault data. This paper involves case studies of NASA projects that can be accessed from the PROMISE repository. Case studies show that CC data help build high-performance fault predictors in the absence of fault labels and remarkable results are achieved. We suggest that companies use CC data if they do not have any historical fault data when they decide to build their fault prediction models.

show abstract

“…Below, a rule of this type can be observed:

R : IF X_{1} = 3 AND X_{2} = Spain THEN Targe t_{italicvalue}

Pairs of variable/interval. These rules are employed by EDER‐SD and GP3‐SD, for example. Next, an instance of this kind of rules is presented:

\begin{matrix} R : italicIF X_{1} & = ([] 1, 3) italicAND \\ X_{2} = italicSpain italicTHEN italicTarge t_{value} \end{matrix}

Pairs of variable/value with order relations.…”

Section: Subgroup Discoverymentioning

confidence: 99%

Overview on evolutionary subgroup discovery: analysis of the suitability and potential of the search performed by evolutionary algorithms

Carmona

González

Jesus

et al. 2014

WIREs Data Min & Knowl

View full text Add to dashboard Cite

Subgroup discovery (SD) is a descriptive data mining technique using supervised learning. In this article, we review the use of evolutionary algorithms (EAs) for SD. In particular, we will focus on the suitability and potential of the search performed by EAs in the development of SD algorithms. Future directions in the use of EAs for SD are also presented in order to show the advantages and benefits that this search strategy contribute to this task.

show abstract

Searching for rules to detect defective modules: A subgroup discovery approach

Cited by 55 publications

References 50 publications

Interpretable Decision Sets

Interpretable Decision Sets

The use of cross-company fault data for the software fault prediction problem

Overview on evolutionary subgroup discovery: analysis of the suitability and potential of the search performed by evolutionary algorithms

Contact Info

Product

Resources

About