Data Science Perspective Research Articles

Fast and Efficient Data Science Techniques for COVID-19 Group Testing

Researchers and public officials tend to agree that until a vaccine is readily available, stopping SARS-CoV-2 transmission is the name of the game. Testing is the key to preventing the spread, especially by asymptomatic individuals. With testing capacity restricted, group testing is an appealing alternative for comprehensive screening and has recently received FDA emergency authorization. This technique tests pools of individual samples, thereby often requiring fewer testing resources while potentially providing multiple folds of speedup. We approach group testing from a data science perspective and offer two contributions. First, we provide an extensive empirical comparison of modern group testing techniques based on simulated data. Second, we propose a simple one-round method based on ${\ell _{1}}$-norm sparse recovery, which outperforms current state-of-the-art approaches at certain disease prevalence rates.

A Method for MBTI Classification Based on Impact of Class Components

Predicting the personality type of text authors has a well-known usage in psychology with practical applications in business. From the data science perspective, we can look at this problem as a text classification task that can be tackled using natural language processing (NLP) and deep learning. This paper proposes a method and a novel loss function for multiclass classification using the Myers–Briggs Type Indicator (MBTI) approach for predicting the author’s personality type. Furthermore, this paper proposes an approach that improves the current results of the MBTI multiclass classification because it considers components of compound class labels as supportive elements for better classification according to MBTI. As such, it also provides a new perspective on this classification problem. The experimental results on long short-term memory (LSTM) and convolutional neural network (CNN) models outperform baseline models for multiclass classification, related research on multiclass classification, and most research with four binary approaches to MBTI classification. Moreover, other classification problems that target compound class labels and label parts with binary mutually exclusive values can benefit from this approach.

Significance, relevance and explainability in the machine learning age: an econometrics and financial data science perspective

Although machine learning is frequently associated with neural networks, it also comprises econometric regression approaches and other statistical techniques whose accuracy enhances with increasing observation. What constitutes high quality machine learning is yet unclear though. Proponents of deep learning (i.e. neural networks) value computational efficiency over human interpretability and tolerate the ‘black box’ appeal of their algorithms, whereas proponents of explainable artificial intelligence (xai) employ traceable ‘white box’ methods (e.g. regressions) to enhance explainability to human decision makers. We extend Brooks et al.’s [2019. ‘Financial Data Science: The Birth of a New Financial Research Paradigm Complementing Econometrics?’ European Journal of Finance 25 (17): 1627–36.] work on significance and relevance as assessment critieria in econometrics and financial data science to contribute to this debate. Specifically, we identify explainability as the Achilles heel of classic machine learning approaches such as neural networks, which are not fully replicable, lack transparency and traceability and therefore do not permit any attempts to establish causal inference. We conclude by suggesting routes for future research to advance the design and efficiency of ‘white box’ algorithms.

A predictive model and country risk assessment for COVID-19: An application of the Limited Failure Population concept.

This article provides predictions for the spread of the SARS-CoV-2 virus for a number of European countries and the United States of America, drawing from their different profiles, both socioeconomically and in terms of outbreak and response to the 2019-2020 coronavirus pandemic, from an engineering and data science perspective. Each country is separately analysed, due to their differences in populations density, cultural habits, health care systems, protective measures, etc. The probabilistic analysis is based on actual data, as provided by the World Health Organization (WHO), as of May 1, 2020. The deployed predictive model provides analytical expressions for the cumulative density function of COVID-19 curve and estimations of the proportion of infected subpopulation for each country. The latter is used to define a Risk Index, towards assessing the level of risk for a country to exhibit high rates of COVID-19 cases after a given interval of observation and given the plans of lifting lockdown measures.

Distributed linguistic representations in decision making: Taxonomy, key elements and applications, and challenges in data science and explainable artificial intelligence

Distributed linguistic representations are powerful tools for modelling the uncertainty and complexity of preference information in linguistic decision making. To provide a comprehensive perspective on the development of distributed linguistic representations in decision making, we present the taxonomy of existing distributed linguistic representations. Then, we review the key elements and applications of distributed linguistic information processing in decision making, including the distance measurement, aggregation methods, distributed linguistic preference relations, and distributed linguistic multiple attribute decision making models. Next, we provide a discussion on ongoing challenges and future research directions from the perspective of data science and explainable artificial intelligence.

How Big Data Confers Market Power to Big Tech: Leveraging the Perspective of Data Science

Data-hungry applications are central to the largest online platforms. Using a novel approach that leverages data science to inform the economics, we demonstrate how data is a source of market power. We highlight the importance of data heterogeneity, whereby small feature differences translate into large value differences. We examine how concept drift, the existence of a nonstationary relationship between the predictive and target variables, implies that access to a continuous stream of data is competitively advantageous. We analyze how an information bottleneck and high sample complexity in existing applications lead to increasing returns to data. Finally, we show how user interaction control enables personalization that raises switching costs. The combined effect is a potent data barrier to entry that endows substantial market power to only the largest online platforms. Competition policy should focus on enabling entrants unfettered access to vast continuous data streams similar to those available to platform incumbents.

How Big Data Confers Market Power to Big Tech: Leveraging the Perspective of Data Science

Data-hungry applications such as targeted advertising, information filtering systems, web search, and virtual assistants have become central to the business models of the largest online platforms. We demonstrate how data acts as a source of market power in these applications using a novel approach that leverages the perspective of data science to inform the economic analysis. We highlight the importance of data heterogeneity, which implies that minor feature differences engender substantial value differences that insulate online platforms from competition. In addition, we call attention to the notion of concept drift, that often in these settings there exists a non-stationary relationship between the predictive variables and the target variable of interest. This signifies that having access to a continuous stream of data is essential to being a viable competitor. Further, we establish that these applications suffer from (i) an information bottleneck, which makes incremental data valuable and at times increasingly so, and (ii) high sample complexity, which mandates the need for vast amounts of clean and rich training data. Finally, we show that online platforms engage in user interaction control that enables the creation of a personalized suite of features that raises switching costs for consumers. The combined effect is to generate significant data barriers to entry that endow the large online platforms with market power. This suggests that policies to facilitate a more competitive landscape in markets involving these applications must focus on enabling access to a continuous stream of high-quality data.

Examining the feasibility of using open data to benchmark building energy usage in cities: A data science and policy perspective

Buildings are by far the largest source of urban energy consumption. In an effort to reduce energy use, cities are mandating that buildings undergo energy benchmarking—the process of measuring building energy performance in order to identify buildings that are inefficient. In this paper, we examine the feasibility of using city-specific, public open data sources in two benchmarking models and compare the results to the same models when using the Commercial Building Energy Consumption Survey (CBECS) dataset, the basis for Energy Star. The two benchmarking models use datasets containing building characteristics and annual energy use from ten major cities. To examine the difference in performance between linear and non-linear models, we use random forest and lasso regression. Results demonstrate that benchmarking models using open data outperform models based solely on the CBECS dataset. Additionally, our results indicate that building area, property type, conditioned area, and water usage are the most important variables for cities to collect. Having demonstrated the benefits of using open data, we recommend two changes to current benchmarking practices: (1) new guidelines that support a data-driven benchmarking framework relying on open data and a transparent modeling process and (2) supporting policies that publicize benchmarking results and incentivize energy savings.

A machine learning approach for predicting computational intensity and domain decomposition in parallel geoprocessing

ABSTRACT High performance computing is required for fast geoprocessing of geospatial big data. Using spatial domains to represent computational intensity (CIT) and domain decomposition for parallelism are prominent strategies when designing parallel geoprocessing applications. Traditional domain decomposition is limited in evaluating the computational intensity, which often results in load imbalance and poor parallel performance. From the data science perspective, machine learning from Artificial Intelligence (AI) shows promise for better CIT evaluation. This paper proposes a machine learning approach for predicting computational intensity, followed by an optimized domain decomposition, which divides the spatial domain into balanced subdivisions based on the predicted CIT to achieve better parallel performance. The approach provides a reference framework on how various machine learning methods including feature selection and model training can be used in predicting computational intensity and optimizing parallel geoprocessing against different cases. Some comparative experiments between the approach and traditional methods were performed using the two cases, DEM generation from point clouds and spatial intersection on vector data. The results not only demonstrate the advantage of the approach, but also provide hints on how traditional GIS computation can be improved by the AI machine learning.

The System Identification and Prediction of the Social Earthquakes Burst in Human Society

The internal mechanism of the social “earthquakes”- the wars burst in human society, is investigated by nonlinear theory. We explore the scaling behavior in the human activity and the historical evolution of ancient China. Based on the war chronology and history chronology from B.C.2000 to A.D.1840, using statistical analysis together with the detrended fluctuation analysis, we find that the social evolution system shows self-organized critical behavior and there exists self-similar scaling behavior during the evolution system. Based on time-delay reconstructed phase space, a modified radial basis function neural network (modified RBF-NN) is used to predict the number of wars and the transformation of the historical stage. It shows that the prediction result matches quite well with the real value for the following two historical stages. Furthermore, by singular value decomposition and the sparse identification algorithm in time-delay coordinates, the coefficients of the candidate functions in identified system can be solved from a sparse regression problem. We extracted the governing equations of big events burst in social system. The identified system fits well with the real data, and it can capture the same topology with the original system of wars. In this paper, we investigated the wars burst in social system by statistic analysis, neural network as well as sparse identification. It sheds light on the study of wars or conflicts from perspective of data science and dynamical theory.

A Cancer Diagnosis Method Combining miRNA-lncRNA Interaction Pairs and Class Weight Competition

From the perspective of data science, we propose a cancer diagnosis method combining miRNA-lncRNA interaction pairs and class weight competition. First, miRNA-lncRNA interaction data is introduced into joint expression profiles, and the complex mechanism of cancer development is demonstrated in depth through the reappearance of key association information. This is an information ensemble of three carcinogenic mechanisms at dataset construction level: classical genetics, epigenetics, and the complex interaction effect between miRNAs and lncRNAs. Then, we put forward a hybrid feature selection algorithm. By preserving the interaction relationship between miRNAs and lncRNAs, it quickly and steadily removes irrelevant and redundant features and solves the high-dimensional disaster problem of cancer expression profiles. This is an information ensemble of multiple feature selection algorithms and the significant association relationship found between multi-dimensional features at feature selection level. A diversity sampling and multi-algorithm learners are used to construct a multiple heterogeneous classification models, which overcomes the small size of normal samples and the local optimum of single algorithm and single mode. This is an information ensemble of multiple classification model structures and multiple classification model state parameters at classification modeling level. At decision level, the proposed class weight which does not depend on the sample size is constructed to address the issue of unbalanced sample class of cancers. The ensemble of multi-category multi-state information at four levels (dataset construction, feature selection, classification modeling, and decision) constitutes the framework of the proposed method. We classify BRCA, LUAD and LUSC in TCGA. Compared with the state-of-the-art classification methods, the proposed method has improved classification accuracy by 9.25%~21.25%, sensitivity by 6.45%~66.45%, and specificity by 10.11%. In addition, we find that lincRNA instead of miRNA always appears in each group of feature genes, which provides a new clue for the locus target selection in cancer treatment.

Deep Fusion for 3D Gaze Estimation From Natural Face Images Using Multi-Stream CNNs

Over the last few decades, eye gaze estimation techniques have been thoroughly investigated by many researchers. However, predicting a 3D gaze from a 2D natural image remains challenging because it has to deal with several issues such as diverse head positions, face shape transformation, illumination variations, and subject individuality. Many previous studies employ convolutional neural networks (CNNs) for this task, and yet the accuracy needs improvement for its practical use. In this paper, we propose a 3D gaze estimation framework based on the data science perspective: First, a novel neural network architecture is designed to exploit every possible visual attribute such as the states of both eyes and the head position, including several augmentations; secondly, the data fusion method is utilized by incorporating multiple gaze datasets. Extensive experiments were carried out using two standard eye gaze datasets, including comparative analysis. The experimental results suggest that our method outperforms state-of-the-art with 2.8 degrees for MPIIGaze and 3.05 degrees for EYEDIAP dataset, respectively, indicating that it has a potential for real applications.

Electric Vehicles: A Data Science Perspective Review

Current trends are showing that the popularity of electric vehicles (EVs) has significantly increased over the last few years, causing changes not only in the transportation industry but generally in business and society. This paper covers one possible angle to the (r) evolution instigated by EVs, i.e., it provides the data science perspective review of the interdisciplinary area at the intersection of green transportation, energy informatics, and economics. Namely, the review summarizes data-driven research in EVs by identifying two main research streams: (i) socio–economic, and (ii) socio–technical. The socio–economic stream includes research in: (i) acceptance of green transportation in countries and among different populations, (ii) current trends in the EV market, and (iii) forecasting future sales for the green transportation. The socio–technical stream includes research in: (i) electric vehicle battery price and capacity and (ii) charging station management. This kind of study is especially important now when the question is no longer whether the transition from internal-combustion engine vehicles to clean-fuel vehicles is going to happen but how fast it will happen and what are going to be implications for society, governmental policies, and industry. Based on the presented literature review, the paper also outlines the most significant open questions and challenges that are yet to be solved: (i) scarcity of trustworthy (open) data, and (ii) designing a generalized methodology for charging station deployment.

Multi-agent system for anomaly detection in Industry 4.0 using Machine Learning techniques

Industry 4.0 is the new industrial stage that is committed to greater automation, connectivity and globalization. The interrelation between the different areas has penetrated the industrial world thanks to the Internet of things and the world of Big Data. This amount of information available in plants is growing increasingly, also aided by the network computing services offered by cloud computing or edge computing. That is why it’s necessary to carry out complex fusion methods and data analysis using Machine Learning techniques to address specific industrial requirements and needs. The central challenge of industry 4.0 from the perspective of data science is to predict the history within the monitored processes, providing as much information as possible, avoiding them and stave off severe economic losses. This article will show a review of the application of Artificial Intelligence (AI) techniques such as Machine Learning (ML) immersed in multi-agent systems (MAS) in Industry 4.0. For this, a bibliographic search has been carried out in databases recognized as Science Direct, Google Scholar, Scopus or Springer, filtering the investigations from 2018 to the actually. The article concludes by pointing out the possible future lines and the importance of the transition towards the implementation of new technologies for the competitiveness of factories.

Understanding Dephosphorization in Basic Oxygen Furnaces (BOFs) Using Data Driven Modeling Techniques

Owing to the continuous deterioration in the quality of iron ore and scrap, there is an increasing focus on improving the Basic Oxygen Furnace (BOF) process to utilize lower grade input materials. The present paper discusses dephosphorization in BOF steelmaking from a data science perspective, which thus enables steelmakers to produce medium and low phosphorus steel grades. In the present study, data from two steel mills (Plant I and Plant II) were collected and various statistical methods were employed to analyze the data. While most operators in steel plants use spreadsheet-based techniques and linear regression to analyze data, this paper discusses on the suitability of selecting various statistical methods, and benchmarking tests to analyze such dephosphorization data sets. The data contains a wide range of operating conditions, both low and high phosphorus input loads, different slag basicity’s, different slag chemistries, and different end point temperatures, etc. The predicted phosphorus partition from various statistical models is compared against plant data and verified against previously published research.

Artificial intelligence in dermato-oncology: a joint clinical and data science perspective.

Artificial intelligence in dermato-oncology: a joint clinical and data science perspective.

Literature review on Big Data and Its Application Fields

In recent years, due to information technology such as computer, Internet and the development of sensor technology, make modern life appeared “everything can be a digital” thinking, as a complex system test, test and condition monitoring of data series in growth, big data is becoming a hot research topic in the field of all kinds of industrial. This paper will start from the background of the era of big data, discuss the features of big data and its application in reality, and elaborate the trend of big data development from the perspective of data science, industry 4.0 and information physics system, and analyze the challenges of big data development.

Understanding Faceted Search from Data Science and Human Factor Perspectives

Faceted search has become a common feature on most search interfaces in e-commerce websites, digital libraries, government’s open information portals, and so on. Beyond the existing studies on developing algorithms for faceted search and empirical studies on facet usage, this study investigated user real-time interactions with facets over the course of a search from both data science and human factor perspectives. It adopted a Random Forest (RF) model to successfully predict facet use using search dynamic variables. In addition, the RF model provided a ranking of variables by their predictive power, which suggests that the search process follows rhythmic flow of a sequence within which facet addition is mostly influenced by its immediately preceding action. In the follow-up user study, we found that participants used facets at critical points from the beginning to end of search sessions. Participants used facets for distinctive reasons at different stages. They also used facets implicitly without applying the facets to their search. Most participants liked the faceted search, although a few participants were concerned about the choice overload introduced by facets. The results of this research can be used to understand information seekers and propose or refine a set of practical design guidelines for faceted search.

Development of Smart Semiconductor Manufacturing: Operations Research and Data Science Perspectives

With advances in information and telecommunication technologies and data-enabled decision making, smart manufacturing can be an essential component of sustainable development. In the era of the smart world, semiconductor industry is one of the few global industries that are in a growth mode to smartness, due to worldwide demand. The important opportunities that can boost the cost reduction of productivity and improve quality in wafer fabrication are based on the simulations of actual environment in Cyber-Physical Space and integrate them with decentralized decision-making systems. However, this integration faced the industry with novel unique challenges. The stream of the data from sensors, robots, and Cyber-Physical Space can aid to make the manufacturing smart. Therefore, it would be an increased need for modeling, optimization, and simulation for the value delivery from manufacturing data. This paper aims to review the success story of smart manufacturing in semiconductor industry with the focus on data-enabled decision making and optimization applications based on operations research and data science perspective. In addition, we will discuss future research directions and new challenges for this industry.

Big Data Market Optimization Pricing Model Based on Data Quality

In recent years, data has become a special kind of information commodity and promoted the development of information commodity economy through distribution. With the development of big data, the data market emerged and provided convenience for data transactions. However, the issues of optimal pricing and data quality allocation in the big data market have not been fully studied yet. In this paper, we proposed a big data market pricing model based on data quality. We first analyzed the dimensional indicators that affect data quality, and a linear evaluation model was established. Then, from the perspective of data science, we analyzed the impact of quality level on big data analysis (i.e., machine learning algorithms) and defined the utility function of data quality. The experimental results in real data sets have shown the applicability of the proposed quality utility function. In addition, we formulated the profit maximization problem and gave theoretical analysis. Finally, the data market can maximize profits through the proposed model illustrated with numerical examples.