Quantifying the sensitivity of seismic facies classification to seismic attribute selection: An explainable machine learning study

During the past two decades, geoscientists have used machine learning to produce a more quantitative reservoir characterization and to discover hidden patterns in their data. However, as the complexity of these models increase, the sensitivity of their results to the choice of the input data becomes more challenging. Measuring how the model uses the input data to perform either a classification or regression task provides an understanding of the data-to-geology relationships which indicates how confident we are in the prediction. To provide such insight, the ML community has developed Local Interpretable Model-agnostic Explanations (LIME), and SHapley Additive exPlanations (SHAP) tools. In this study, we train a random forest architecture using a suite of seismic attributes as input to differentiate between mass transport deposits (MTDs), salt, and conformal siliciclastic sediments in a Gulf of Mexico dataset. We apply SHAP to understand how the model uses the input seismic attributes to identify target seismic facies and examine in what manner variations in the input such as adding band-limited random noise or applying a Kuwahara filter impact the models’ predictions. During our global analysis, we find that the attribute importance is dynamic, and changes based on the quality of the seismic attributes and the seismic facies analyzed. For our data volume and target facies, attributes measuring changes in dip and energy show the largest importance for all cases in our sensitivity analysis. We note that to discriminate between the seismic facies, the ML architecture learns a “set of rules” in multi-attribute space and that overlap between MTDs, salt, and conformal sediments might exist based on the seismic attribute analyzed. Finally, using SHAP at a voxel-scale, we understand why certain areas of interest were misclassified by the algorithm and perform an in-context interpretation to analyze how changes in the geology impact the model’s predictions.

Intervention Characteristics Considered in Health Educators’ Adoption Decision-Making Process

It is well known that perceptions of intervention characteristics (e.g., cost, source, evidence strength and quality) are a critical link from dissemination to implementation. What is less known is the process by which researchers understand the characteristics most valued by key intermediaries (i.e., real-world decision-makers), particularly in the federal system of Cooperative Extension. In Extension, university-based specialists are available to assist county-based agents in program selection, delivery, and evaluation. For this work, a sequential explanatory mixed-methods design was used to conduct surveys and semi-structured interviews, informed by the Diffusion of Innovations theory and Consolidated Framework for Implementation Research. Educators and specialists were recruited across 47 states to identify characteristics of health promotion interventions that facilitate the adoption decision-making process. Analysis of intervention attribute importance survey data was conducted through a one-way ANOVA with Bonferroni post hoc test to determine individual variable differences between responses. Interviews underwent a conventional content analysis. In total, 121 educators and 47 specialists from 33 states completed the survey. Eighteen educators and 10 specialists completed interviews. Educators and specialists valued components such as the community need for the intervention, and potential reach compared with other components including previous delivery settings and external funding of the intervention ( p < .05). Qualitative data indicated divergence between educators and specialists; educators valued understanding the intervention cost (time and training) and specialists valued the evidence base and external funding available. Intervention developers should communicate information valued by different stakeholders to improve the adoption of evidence-based interventions.

Decision Tool of Medical Endoscope Maintenance Service in Chinese Hospitals: A Conjoint Analysis

Background: Quality and cost of medical device maintenance are dominant factors influencing hospitals’ decision in choosing medical endoscope products. Effective and efficient medical device maintenance are also paramount for providing cost-effective and high quality of medical care. This research aims to facilitate decision-making at hospitals in choosing the suitable endoscope device and the associated maintenance service; it also aims to facilitate decision-making at suppliers to develop the right products and services to fulfill customer demands.Methods: A cross-sectional survey was undertaken in 50 Chinese hospitals, including primary and tertiary hospitals. Moreover, 65 medical staff and 56 medical engineers were recruited from 50 Chinese hospitals. A comprehensive set of attributes were defined and investigated. Conjoint analysis and orthogonal design were used for survey design and statistical analysis. Results: Attribute importance and utility values of decision-making factors were analyzed at the aggregate, occupation, and medical institution levels. (1) At the aggregate level, the most critical factor is "maintenance response" and the least important one is "maintenance efficiency". (2) At the occupation level, medical staff paid more attention to "maintenance response" and medical engineers paid more attention to "maintenance quality". (3) At the medical institution level, Primary hospitals paid more attention to "maintenance price", while tertiary hospitals paid more attention to "maintenance quality". Conclusions: In general, this study provides a more scientific decision-making tool to both hospitals in choosing maintenance service of medical equipment such as endoscopy, and it also helps manufacturers and suppliers improve the after-sales service.

Literature in EFL classes: four teachers' experiences in Brazil

Some of the benefits of using literature in EFL classes include improving vocabulary, critical thinking and providing authentic material. Researchers such as Melo Júnior (2015) and Lago (2016; 2017) investigate literature in EFL contexts, noticing how literature can also motivate students. This paper aims to analyze English teachers' experience in Goiás, Brazil, and their relationship with literature in their EFL classes. Its purpose is to investigate whether these teachers use literary works during their lessons. Four teachers were interviewed for this case study. The results show these teachers attribute importance to literature in EFL contexts.

Prediction of the effective reproduction number of COVID-19 in Greece. A machine learning approach using Google mobility data

This paper demonstrates how a short-term prediction of the effective reproduction number (Rt) of COVID-19 in regions of Greece is achieved based on online mobility data. Various machine learning methods are applied to predict Rt and attribute importance analysis is performed to reveal the most important variables that affect the accurate prediction of Rt. Work and Park categories are identified as the most important mobility features when compared to the other attributes, with values of 0.25 and 0.24, respectively. Our results are based on an ensemble of diverse Rt methodologies to provide non-precautious and non-indulgent predictions. Random Forest algorithm achieved the highest R2 (0.8 approximately), Pearson’s and Spearman’s correlation values close to 0.9, outperforming in all metrics the other models. The model demonstrates robust results and the methodology overall represents a promising approach towards COVID-19 outbreak prediction. This paper can help health-related authorities when deciding on non-nosocomial interventions to prevent the spread of COVID-19.

A Supervised Machine-Learning Prediction of Textile’s Antimicrobial Capacity Coated with Nanomaterials

Textile materials, due to their large surface area and moisture retention capacity, allow the growth of microorganisms, causing undesired effects on the textile and on the end-users. The textile industry employs nanomaterials (NMs)/composites and nanofibers to enhance textile features such as water/dirt-repellent, conductivity, antistatic properties, and enhanced antimicrobial properties. As a result, textiles with antimicrobial properties are an area of interest to both manufacturers and researchers. In this study, we present novel regression models that predict the antimicrobial activity of nano-textiles after several washes. Data were compiled following a literature review, and variables related to the final product, such as the experimental conditions of nano-coating (finishing technologies) and the type of fabric, the physicochemical (p-chem) properties of NMs, and exposure variables, were extracted manually. The random forest model successfully predicted the antimicrobial activity with encouraging results of up to 70% coefficient of determination. Attribute importance analysis revealed that the type of NM, shape, and method of application are the primary features affecting the antimicrobial capacity prediction. This tool helps scientists to predict the antimicrobial activity of nano-textiles based on p-chem properties and experimental conditions. In addition, the tool can be a helpful part of a wider framework, such as the prediction of products functionality embedded into a safe by design paradigm, where products’ toxicity is minimized, and functionality is maximized.

Analysis of Consumer’s Preference on Purchasing Fresh Fish (Case Study in Muara Baru Modern Fish Market, DKI Jakarta)

The fisheries sector is one of the agricultural sub-sectors that has a role in providing food for the people of Indonesia. Consumers have behavior in purchasing fish in meeting their needs or desires to obtain a product. This study aims to analyze consumer preferences for the type of fish and the order of attributes. The method used in this research is a case study. The research location was carried out at the Muara Baru Modern Fish Market (PIM) DKI Jakarta. The data used are primary data and secondary data. The primary data collection technique used accidental sampling with a sample of 100 respondents while the secondary data were obtained from Perum Perikanan Indonesia as the manager of the Muara Baru Modern Fish Market, the Library, the Central Statistics Agency (BPS) DKI Jakarta, the National Statistics Agency (BPS). Consumer preference analysis used attitude measurement analysis measured by Chi-square and based on rank orders analysis to determine the order of attribute importance. Based on the study results, it was shown that all the attributes observed in this study were significantly different at the 95% confidence level. In contrast, the analysis of the level of importance of the attributes showed that the priority of consumers' interests in fish in the Muara Baru Modern Fish Market (PIM) was price, freshness, cleanliness, texture, and fish scent.

Incorporating heterogeneity of travelers’ preferences into the overall hostel performance rating

Hostels have become a very popular form of accommodation and their varieties have grown steadily in recent years. To ensure the sustainability of this business model, it is necessary to understand the main drivers influencing travelers to choose a hostel accommodation. For this purpose, we conducted an online survey using convenience sampling and purposive sampling techniques. Respondents' preferences to six hostel attributes (cleanliness, location, staff, atmosphere, facilities, and cancellation policy) were determined using discrete choice analysis. Sample results showed that the most important attributes are cleanliness and location, while the atmosphere is the least important one. However, widespread heterogeneity in preferences was observed, and cluster analyzes identified three distinct groups of travelers: “cleanliness sticklers”, “location demanders” and “party seekers”. Facilities and atmosphere were found to be very important attributes for particular clusters. These findings can help design a marketing strategy for each of the identified segments to ensure sustainable business. Finally, we have proposed a new approach to calculating the hostel overall rating based on attribute importance, which shows much better discriminatory power compared to the traditional average-based approach.

A Novel Activation Maximization-based Approach for Insight into Electrophysiology Classifiers

Spectral analysis remains a hallmark approach for gaining insight into electrophysiology modalities like electroencephalography (EEG). As the field of deep learning has progressed, more studies have begun to train deep learning classifiers on raw EEG data, which presents unique problems for explainability. A growing number of studies have presented explainability approaches that provide insight into the spectral features learned by deep learning classifiers. However, existing approaches only attribute importance to different frequency bands. Most of the methods cannot provide insight into the actual spectral values or the relationship between spectral features that models have learned. Here, we present a novel adaptation of activation maximization for electrophysiology time-series that generates samples that indicate the features learned by classifiers by optimizing their spectral content. We evaluate our approach within the context of EEG sleep stage classification with a convolutional neural network, and we find that our approach is able to identify spectral patterns known to be associated with each sleep stage. We also find surprising results suggesting that our classifier may have prioritized the use of eye and motion artifact when identifying Awake samples. Our approach is the first adaptation of activation maximization to the domain of raw electrophysiology classification. Additionally, our approach has implications for explaining any classifier trained on highly dynamic, long time-series.

EXPRESS: Attribute Sentiment Scoring with Online Text Reviews: Accounting for Language Structure and Missing Attributes

The authors address two significant challenges in using online text reviews to obtain finegrained attribute level sentiment ratings. First, in contrast to methods that rely on word frequency, they develop a deep learning convolutional-LSTM hybrid model to account for language structure. The convolutional layer accounts for spatial structure (adjacent word groups or phrases) and LSTM accounts for sequential structure of language (sentiment distributed and modified across non-adjacent phrases). Second, they address the problem of missing attributes in text in constructing attribute sentiment scores—as reviewers write only about a subset of attributes and remain silent on others. They develop a model-based imputation strategy using a structural model of heterogeneous rating behavior. Using Yelp restaurant review data, they show superior attribute sentiment scoring accuracy with their model. They find three reviewer segments with different motivations: status seeking, altruism/want voice, and need to vent/praise. Reviewers write to inform and vent/praise, but not based on attribute importance. The heterogeneous model-based imputation performs better than other common imputations; and importantly leads to managerially significant corrections in restaurant attribute ratings. More broadly, our results suggest that social science research should pay more attention to reduce measurement error in variables constructed from text.