Variable importance (VI) tools describe how much covariates contribute to a prediction model’s accuracy. However, important variables for one well-performing model (for example, a linear model
f
(x) ...= x
T
β
with a fixed coefficient vector
β
) may be unimportant for another model. In this paper, we propose model class reliance (MCR) as the range of VI values across
all
well-performing model in a prespecified class. Thus, MCR gives a more comprehensive description of importance by accounting for the fact that many prediction models, possibly of different parametric forms, may fit the data well. In the process of deriving MCR, we show several informative results for permutation-based VI estimates, based on the VI measures used in Random Forests. Specifically, we derive connections between permutation importance estimates for a
single
prediction model, U-statistics, conditional variable importance, conditional causal effects, and linear model coefficients. We then give probabilistic bounds for MCR, using a novel, generalizable technique. We apply MCR to a public data set of Broward County criminal records to study the reliance of recidivism prediction models on sex and race. In this application, MCR can be used to help inform VI for unknown, proprietary models.
Abstract Concerns with taste, nutrition, cost, and convenience are said to be key influences on food choices. This study examined the importance of food-related attitudes in relation to diet quality ...using US national level data. Interactions by socioeconomic status (SES), gender and race/ethnicity were tested. Analyses of 8957 adults from National Health and Nutrition Examination Survey (NHANES 2007–2010) were conducted in 2014–15. Perceived importance of taste, nutrition, cost, and convenience in dietary choices were assessed using 4-point Likert scales. Education and family income-to-poverty ratio (FIPR) were SES indicators. Healthy Eating Index (HEI-2010), a measure of adherence to 2010 dietary guidelines, was the diet quality measure. Survey-weighted regressions examined associations between attitudes and HEI, and tested for interactions. Taste was rated as "very important" by 77.0% of the US adults, followed by nutrition (59.9%), cost (39.9%), and convenience (29.8%). However, it was the perceived importance of nutrition that most strongly predicted HEI (β: + 8.0 HEI scores among "very important" vs. "not at all important"). By contrast, greater importance for taste and convenience had a weak inverse relation with HEI (β: − 5.1 and − 1.5 respectively), adjusting for SES. Significant interactions were observed by race/ethnicity, but not SES and gender. Those who prioritize nutrition during food shopping achieve higher-quality diets regardless of gender, education and income in the US. Certain racial/ethnic groups manage to eat healthy despite attaching importance to cost and convenience. This is the first evidence of nutrition resilience among US adults using national data, which has huge implications for nutrition interventions.
► Introducing an importance measure. ► Considering two scenarios. ► Considering extensions.
In reliability engineering, component importance measures are used to prioritise components in a system for ...purposes such as reliability improvement and maintenance planning. Existing importance measures have paid little attention to the costs incurred by maintaining a system and its components within a given time period. Cost-effectiveness analysis, however, is critically important in increasingly competitive markets. This paper proposes a new cost-based importance measure which considers costs incurred by maintaining a system and its components within a finite time horizon. Possible extensions are discussed and examples are given to show the use of the new measure.
The freshwater Indian apple snail, Pila globosa (Swainson, 1822), is well adapted to the equatorial and tropical regions of the planet, where there are periods of heavy rain that are followed by dry ...spells. It is the most important biotic component of the ecosystem and a dominating member of its communities, making it crucial for the health of the ecosystem. It has a significant economic importance in the international trade market. The flesh of P. globosa is used in aquaculture and as a human protein supplement because of its high protein and low fat content, along with essential fatty acids. The shell of P. globosa is a good source of minerals, especially calcium. P. globosa has been employed in traditional medicinal practices to treat diseases like high blood pressure, heart disease, asthma, rickets, rheumatoid arthritis, osteoporosis, calcium metabolism, bleeding piles, constipation, diarrhoea, smallpox, syphilis, dizziness, anxiety, nervousness, urticaria, night blindness, and conjunctivitis. It is also used to regulate body temperature, to speed up wound healing, to treat circulatory issues, to revive virility and vitality, to treat weakness, and for vision improvement. P. globosa has antimicrobial, anti-inflammatory, anti-oxidative, anti-cancer, and immune boosting properties, and it can be a benefit to mankind. This review provides an overview of the ecological and economic importance, nutritional and ethno-medicinal values of the snail P. globosa.
Measuring variable importance for computational models or measured data is an important task in many applications. It has drawn our attention that the variable importance analysis (VIA) techniques ...were developed independently in many disciplines. We are strongly aware of the necessity to aggregate all the good practices in each discipline, and compare the relative merits of each method, so as to instruct the practitioners to choose the optimal methods to meet different analysis purposes, and to guide current research on VIA. To this end, all the good practices, including seven groups of methods, i.e., the difference-based variable importance measures (VIMs), parametric regression and related VIMs, nonparametric regression techniques, hypothesis test techniques, variance-based VIMs, moment-independent VIMs and graphic VIMs, are reviewed and compared with a numerical test example set in two situations (independent and dependent cases). For ease of use, the recommendations are provided for different types of applications, and packages as well as software for implementing these VIA techniques are collected. Prospects for future study of VIA techniques are also proposed.
•All the good practices for variable importance analysis (VIA) are reviewed.•These VIA techniques are compared with a numerical example.•The relative merit of each technique is analyzed.•Recommendations are provided for different types of applications.•Packages and software for performing these techniques are collected and summarized.
Many importance measures have been proposed with respect to the diverse considerations of system performance, reflecting different probabilistic interpretations and potential applications. This paper ...studies importance measures in reliability, including their definitions, probabilistic interpretations, properties, computations, and comparability. It categorizes importance measures into the structure, reliability, and lifetime types based on the knowledge for determining them. It covers importance measures of individual components, and ones of pairs and groups of components. It also investigates importance measures in consecutive- k -out-of- n systems.
In reliability engineering, time on performing preventive maintenance (PM) on a component in a system may affect system availability if system operation needs stopping for PM. To avoid such an ...availability reduction, one may adopt the following method: if a component fails, PM is carried out on a number of the other components while the failed component is being repaired. This ensures PM does not take system׳s operating time. However, this raises a question: which components should be selected for PM? This paper introduces an importance measure, called component maintenance priority (CMP), that is used to select components for PM. The paper then compares the CMP with other importance measures and studies the properties of the CMP. Numerical examples are given to show the validity of the CMP.
•Introduced an importance measure for prioritising units for preventive maintenance.•Investigated the relationships between the new measure with other existing measures.•Derived the lower and upper bounds of the number of failures for a set of units.
The effectiveness of contemporary strategies for conducting fault tree/event tree (FTET) analyses within the realm of probabilistic risk assessment has recently come under rigorous examination. In ...light of such investigation, facility managers have gained a more profound understanding of the risk and safety implications inherent in the structural and componential integrity of systems (structures, systems, and components). This comprehensive research endeavor harnesses the power of risk models, employing both FTET and binary decision diagrams, to scrutinize and optimize the operational performance of a 10-MW reference Russian research reactor Water-Water Research Reactor (VVR) within the framework of probabilistic safety assessment. Moreover, this investigation delves into the intricate web of interrelationships existing among an array of analytical methodologies. These encompass the Fussell-Vesely (FV) importance measure, criticality analysis, Birnbaum analysis, risk achievement worth (RAW), and the differential importance measure, all with a focus on specific foundational events and vital components. Additionally, this note delves into the analysis of multiple significant measures frequently employed for VVR. Notably, the study establishes that merely two importance measures (IMs) prove sufficient for the core damage equation. Furthermore, this note investigates various important measures often employed for VVR. It is shown that two IMs are enough for the core damage equation. In conclusion, RAW, FV importance, or a blend of the two are adequate enough to be frequently employed for the VVR.
Convolutional Neural Networks (CNNs) have demonstrated outstanding performance in various domains, such as face recognition, object detection, and image segmentation. However, the lack of ...transparency and limited interpretability inherent in CNNs pose challenges in fields such as medical diagnosis, autonomous driving, finance, and military applications. Several studies have explored the interpretability of CNNs and proposed various post-hoc interpretable methods. The majority of these methods are feature-based, focusing on the influence of input variables on outputs. Few methods undertake the analysis of parameters in CNNs and their overall structure. To explore the structure of CNNs and intuitively comprehend the role of their internal parameters, we propose an Attribution Graph-based Interpretable method for CNNs (AGIC) which models the overall structure of CNNs as graphs and provides interpretability from global and local perspectives. The runtime parameters of CNNs and feature maps of each image sample are applied to construct attribution graphs (At-GCs), where the convolutional kernels are represented as nodes and the SHAP values between kernel outputs are assigned as edges. These At-GCs are then employed to pretrain a newly designed heterogeneous graph encoder based on Deep Graph Infomax (DGI). To comprehensively delve into the overall structure of CNNs, the pretrained encoder is used for two types of interpretable tasks: (1) a classifier is attached to the pretrained encoder for the classification of At-GCs, revealing the dependency of At-GC’s topological characteristics on the image sample categories, and (2) a scoring aggregation (SA) network is constructed to assess the importance of each node in At-GCs, thus reflecting the relative importance of kernels in CNNs. The experimental results indicate that the topological characteristics of At-GC exhibit a dependency on the sample category used in its construction, which reveals that kernels in CNNs show distinct combined activation patterns for processing different image categories, meanwhile, the kernels that receive high scores from SA network are crucial for feature extraction, whereas low-scoring kernels can be pruned without affecting model performance, thereby enhancing the interpretability of CNNs.
•A graph construction method is proposed to model CNNs as attribution graphs (At-GCs).•A classifier is built to reveal distinct kernel activation patterns across image categories in CNN.•A scoring network is built to assess node importance in At-GCs, mirroring kernel roles in CNN.