Field Of Data Science Research Articles

Librarian’s perception on data librarianship core concepts: a survey of motivational factors, challenges, skills and appropriate trainings platforms

Purpose Data librarianship, or data-driven librarianship, is the combination of information science, data science and e-science fields and is gaining gradual importance in the library and information science (LIS) profession. Hence, this study investigates the data librarianship core concepts (motivational factors, challenges, skills and appropriate training platforms) to learn and successfully launch data librarianship services.Design/methodology/approach A survey method was used and the data were collected through online questionnaire. Purposive sampling method was applied and 132 responses were received with 76 respondents from the public and 56 from the private sector universities of Pakistan. The statistical package for social sciences (SPSS version 25) was used, and descriptive and inferential statistics were applied to analyzed the data.Findings LIS professionals understand the importance of data-driven library services and perceive that such services are helpful in evolving the image of the library, helping with the establishment of institutional data repositories/data banks, developing data resources and services for library patrons and especially researchers, and receiving appreciation and acknowledgment from the higher authorities. The major challenges that emerged from the data were: missing data policies, limited training opportunities for data librarianship roles, no additional financial benefits, lack of infrastructure and systems, lack of organizational support for the initiation of data-driven services, and lack of skills, knowledge and expertise. Data librarianship is in its early stages in Pakistan, and consequently, the LIS professionals are lacking basic, advanced and technical data-driven skills.Research limitations/implications The policy, theoretical and practical implications describe an immediate need for framing data policies. Such policies will help the libraries or any other relevant entities to store the data and assign metadata and documentation in such a way that it is easy to retrieve and reusable for others.Originality/value This is the first study in Pakistan to investigate the perceptions of LIS professionals about data librarianship core concepts: motivational factors, challenges, skills and appropriate training platforms to grasp data-driven skills and successfully launch library services.

Comparative analysis of resampling techniques on Machine Learning algorithm

Generally, classification algorithms in the field of data science assume that the classes of training data are equally distributed. However, datasets on real problems often have an unbalanced class distribution. Unbalanced dataset classes make up the majority class and the minority class. In general, minority classes are more attractive and more important to identify. In this case, the correct classification for the minority class sample is more valuable than the majority class. The unbalanced class distribution causes the classification algorithm to have difficulty in classifying minority class samples correctly. If the performance of the algorithm model is good for the majority class sample but bad for the minority class then this imbalance problem is a crucial thing to be addressed. Many solutions are offered for this problem, namely by oversampling techniques in the minority class and/or undersampling techniques in the majority class. In this study, the authors tried various sampling techniques and tested them on various machine learning classification algorithms to find out the combination of resampling techniques and algorithms that have high recall in classifying minority class samples and still considering the majority class classification.

Spatial-temporal data analysis of digital twin

Background: Digital Twin (DT) has proven to be one of the most promising technologies for routine monitoring and management of complex systems with uncertainties. Methods: Our work, which is mainly concerned with heterogeneous spatial-temporal data, focuses on exploring data utilization methodology in DT. The goal of this research is to summarize the best practices that make the spatial-temporal data analytically tractable in a systematic and quantifiable manner. Some methods are found to handle those data via jointly spatial-temporal analysis in a high-dimensional space effectively. We provide a concise yet comprehensive tutorial on spatial-temporal analysis considering data, theories, algorithms, indicators, and applications. The advantages of our spatial-temporal analysis are discussed, including model-free mode, solid theoretical foundation, and robustness against ubiquitous uncertainty and partial data error. Finally, we take the condition-based maintenance of a real digital substation in China as an example to verify our proposed spatial-temporal analysis mode. Results: Our proposed spatial-temporal data analysis mode successfully turned raw chromatographic data, which are valueless in low-dimensional space, into an informative high-dimensional indicator. The designed high-dimensional indicator could capture the ’insulation’ correlation among the sampling data over a long time span. Hence it is robust against external noise, and may support decision-making. This analysis is also adaptive to other daily spatial-temporal data in the same form. Conclusions: This exploration and summary of spatial-temporal data analysis may benefit the fields of both engineering and data science.

The Modern World Landscape of Data Science Online Education

Data science as an emerging branch of applied knowledge and a new field of study is showing a strong momentum. Besides, the corresponding sphere of educational research is actively developing. At the same time, most of the scientific publications are aimed at studying specific issues related to the content of the programs and their methodological support. The wider context and especially the international perspective are lacking for the necessary attention of researchers.In this regard, the purpose of our study was to summarize and systematize information about training programs in the field of data science presented on online platforms of the main macro-regions – America, Europe and Asia. For this purpose, we found out what elements the corpus of data science training programs consists of, as well as how courses are distributed on educational platforms by countries, organizational providers, level of education and duration of study. Based on the data obtained, we conducted a comparative interregional study of educational programs presented on online platforms.The findings made it possible to draw conclusions about the specifics of the global landscape of data science online education, as well as to determine the specifics of the Russian segment and to formulate recommendations for solving significant problems of the domestic economy using data science online education.

A Novel 2D Clustering Algorithm Based on Recursive Topological Data Structure

In the field of data science and data mining, the problem associated with clustering features and determining its optimum number is still under research consideration. This paper presents a new 2D clustering algorithm based on a mathematical topological theory that uses a pseudometric space and takes into account the local and global topological properties of the data to be clustered. Taking into account cluster symmetry property, from a metric and mathematical-topological point of view, the analysis was carried out only in the positive region, reducing the number of calculations in the clustering process. The new clustering theory is inspired by the thermodynamics principle of energy. Thus, both topologies are recursively taken into account. The proposed model is based on the interaction of particles defined through measuring homogeneous-energy criterion. Based on the energy concept, both general and local topologies are taken into account for clustering. The effect of the integration of a new element into the cluster on homogeneous-energy criterion is analyzed. If the new element does not alter the homogeneous-energy of a group, then it is added; otherwise, a new cluster is created. The mathematical-topological theory and the results of its application on public benchmark datasets are presented.

The Emerging Field of Medical Regulatory Technology and Data Science

Regulations contain rules setup by (governmental) authorities to control specific aspects of certain industries, which often influences the way companies operate [...]

Importance of statistics to data science

This paper is mainly discussed on importance and contribution of statistics to Data science and how it emerges as the most important factor to solve realistic problems which contains huge amount of data processing. There are various methods in statistics which help Analysis in data science which will be explained in detail. This work also emphasizes on importance of Data Science in this present technology. Statistics is proved to be an important discipline in regulating the work analyzed in the field of Data Science. This work compare various statistical approaches with This outlines the numerous potential data analysis approach processes which helps in examining the influence of quantitative statistical measures on data collection and optimization, data interpretation, data processing and modelling, testing and presenting and Various challenges faced in the process of data science using statistics is given in brief. Here there is a numerous way to enhance the data science techniques with the help of statistics methodologies.

Reliable Algorithms for Machine Learning Models: Implementation Research in Data Science

Machine Learning generates programs that make predictions and informed decisions about complex problems in an efficient and reliable way. These ML programs autonomously test solutions against the dataset to find the best fit for the problem. The paper aims to review the ML algorithms that develop prediction models by utilizing training dataset and known output. The paper also focuses on ML principles, algorithms, approaches, and applications for Supervised, Unsupervised, and Reinforcement learning that can perform tasks without being explicitly programmed for it. Completely opposite to rule-based programming, the machine learning paradigm uses examples of real data sets and pre-process it before providing the desired outputs based on these examples. In the case of more involved and complex tasks, it can be challenging for humans to explicitly program the models. On the other hand, it can be more effective to help the machines develop the algorithms for advanced tasks. This paper will also present the trending real-world applications of Machine Learning in Image Recognition and Biomedicine. Additionally, it will provide a background analysis of machine learning and related fields of data science.

Identification of Digital Health Priorities for Palliative Care Research: Modified Delphi Study.

BackgroundDevelopments in digital health have the potential to transform the delivery of health and social care to help citizens manage their health. Currently, there is a lack of consensus about digital health research priorities in palliative care and a lack of theories about how these technologies might improve care outcomes. Therefore, it is important for health care leaders to identify innovations to ensure that an increasingly frail population has appropriate access to palliative care services. Consequently, it is important to articulate research priorities as the first step in determining how finite resources should be allocated to a field saturated with rapidly developing innovation.ObjectiveThe aim of this study is to identify research priority areas for digital health in palliative care.MethodsWe selected digital health trends, most relevant to palliative care, from a list of emerging trends reported by a leading institute of quantitative futurists. We conducted 2 rounds of the Delphi questionnaire, followed by a consensus meeting and public engagement workshop to establish a final consensus on research priorities for digital technology in palliative care. We used the views of public representatives to gain their perspectives on the agreed priorities.ResultsA total of 103 experts (representing 11 countries) participated in the first Delphi round. Of the 103 experts, 55 (53.3%) participated in the second round. The final consensus meetings were attended by 10.7% (11/103) of the experts. We identified 16 priority areas, which involved many applications of technologies, including care for patients and caregivers, self-management and reporting of diseases, education and training, communication, care coordination, and research methodology. We summarized the priority areas into eight topics: big data, mobile devices, telehealth and telemedicine, virtual reality, artificial intelligence, smart home, biotechnology, and digital legacy.ConclusionsThe priorities identified in this study represent a wide range of important emerging areas in the fields of digital health, personalized medicine, and data science. Human-centered design and robust governance systems should be considered in future research. It is important that the risks of using these technologies in palliative care are properly addressed to ensure that these tools are used meaningfully, wisely, and safely and do not cause unintentional harm.

LSOP: Layer-Scaled One-shot Pruning

Neural network pruning is a technique that removes unnecessary weight parameters from a network to decrease its memory and computational requirements. Many different pruning techniques have been proposed to reduce networks with over 90% shrinkage in size while minimizing accuracy loss. This paper aims to establish a framework that can generalize the mechanism among various pruning techniques, which can be used to guide users to design better deep pruning methods in the future. With some basic concepts and findings from data matrix approximation, the framework can explain the success of the state-of-the-art methods as well as a more generalized pruning design (Layer-Scaled One-shot Pruning, LSOP) proposed in this work. After pruning with different algorithms and measuring their accuracies, the researcher also found that those methods or algorithms aligned with the proposed framework were more accurate at sparser networks (density < 10%) than the methods that did not. This suggests that future research into neural network pruning can focus on the proposed framework, which has the potential to accelerate the development of pruning technology and adoption of more efficient neural networks.
 
 The LSOP framework’s capability to explain strong pruning performances implies that polynomial decay and Low-Rank Matrix Approximation techniques from the field of data science can provide support for neural network pruning.

Statistical Analysis of the People Fully Vaccinated against COVID-19 in Two Different Regions.

Motivation. Currently, the COVID-19 pandemic represents a critical issue all over the world. On May 11, 2020, at 05 : 41 GMT, approximately 0.28 million individuals had perished because of the COVID-19 pandemic, and the figure is continuously growing rapidly. Unfortunately, millions of people have died due to this pandemic. As a result, this issue forced governments and other corresponding organizations to take significant action, such as the lockdown and vaccinations. Furthermore, scientists have developed several vaccinations, and the World Health Organization (WHO) has urged governments and people to get vaccinated to eradicate this pandemic. Consequently, the findings of any scientific research into this phenomenon are highly interesting. Problem Statement. To enhance individual protection, it is now critical to analyze and compare the percentage of people fully vaccinated against COVID-19. It is constantly of interest in the field of big data science and other related disciplines to provide the best analysis and modeling of COVID-19 data. Methodology. Through this paper, we aimed to compare individuals who have been completely vaccinated against COVID-19 in two locations: North American countries and Arabian Peninsula countries. Simple techniques for comparing individuals who have been completely vaccinated against COVID-19 have been applied, which may be used to generate the foundation for conclusions. Most significantly, a modern statistical model was created to present the best assessment of individuals completely vaccinated against COVID-19 data in nations in North America and the Arabian Peninsula. Some of the suggested statistical model features were proposed. Furthermore, the estimate of the model parameters was driven using the maximum likelihood estimation method. Results. The flexibility provided by the proposed statistical model is useful for describing the percentage of the individuals completely vaccinated against COVID-19, which provides a close fit with the COVID-19 data. Implications. The proposed statistical model can be used for statistics and generate new statistical distributions that can be used to compare and predict the process of people's willingness to vaccinate and take the vaccine to try to eliminate COVID-19.

Generative Adversarial Networks for Data Generation in Structural Health Monitoring

Structural Health Monitoring (SHM) has been continuously benefiting from the advancements in the field of data science. Various types of Artificial Intelligence (AI) methods have been utilized to assess and evaluate civil structures. In AI, Machine Learning (ML) and Deep Learning (DL) algorithms require plenty of datasets to train; particularly, the more data DL models are trained with, the better output it yields. Yet, in SHM applications, collecting data from civil structures through sensors is expensive and obtaining useful data (damage associated data) is challenging. In this paper, one-dimensional (1-D) Wasserstein loss Deep Convolutional Generative Adversarial Networks using Gradient Penalty (1-D WDCGAN-GP) is utilized to generate damage-associated vibration datasets that are similar to the input. For the purpose of vibration-based damage diagnostics, a 1-D Deep Convolutional Neural Network (1-D DCNN) is built, trained, and tested on both real and generated datasets. The classification results from the 1-D DCNN on both datasets resulted in being very similar to each other. The presented work in this paper shows that, for the cases of insufficient data in DL or ML-based damage diagnostics, 1-D WDCGAN-GP can successfully generate data for the model to be trained on.

Surgical data science - from concepts toward clinical translation.

Recent developments in data science in general and machine learning in particular have transformed the way experts envision the future of surgery. Surgical Data Science (SDS) is a new research field that aims to improve the quality of interventional healthcare through the capture, organization, analysis and modeling of data. While an increasing number of data-driven approaches and clinical applications have been studied in the fields of radiological and clinical data science, translational success stories are still lacking in surgery. In this publication, we shed light on the underlying reasons and provide a roadmap for future advances in the field. Based on an international workshop involving leading researchers in the field of SDS, we review current practice, key achievements and initiatives as well as available standards and tools for a number of topics relevant to the field, namely (1) infrastructure for data acquisition, storage and access in the presence of regulatory constraints, (2) data annotation and sharing and (3) data analytics. We further complement this technical perspective with (4) a review of currently available SDS products and the translational progress from academia and (5) a roadmap for faster clinical translation and exploitation of the full potential of SDS, based on an international multi-round Delphi process.

Estimating Density Ratio of Marginals to Joint: Applications to Causal Inference

In various fields of data science, researchers often face problems of estimating the ratios of two probability densities. Particularly in the context of causal inference, the product of marginals for a treatment variable and covariates to their joint density ratio typically emerges in the process of constructing causal effect estimators. This article applies the general least square density ratio estimation methodology by Kanamori, Hido and Sugiyama to the product of marginals to joint density ratio, and demonstrates its usefulness particularly for causal inference on continuous treatment effects and dose-response curves. The proposed method is illustrated by a simulation study and an empirical example to investigate the treatment effect of political advertisements in the U.S. presidential campaign data.

Theoretical Investigation on Non-Covalent Interactions

This editorial is dedicated to announcing the Special Issue “Theoretical investigation on non-covalent interactions” of Crystals. The Special Issue covers the most recent progress in the rapidly growing fields of data science, artificial intelligence, and quantum and computational chemistry in topics relevant to the problem of theoretical investigation on non-covalent interactions (including, but not limited to, hydrogen, halogen, chalcogen, pnictogen, tetrel, and semi-coordination bonds; agosic and anagosic interactions; stacking, anion-/cation–π interactions; metallophilic interactions, etc.). The main successes of my colleagues and I in the field of fundamental theoretical studies of non-covalent interactions in various chemical compounds over the past year are briefly highlighted.

Ridge Regression based Missing Data Estimation with Dimensionality Reduction: Microarray Gene Expression Data

Data is considered to be the important element in the field of Data Science and Machine Learning. Performance of Machine Learning and Data Mining algorithms greatly influenced by the characteristics of data and data with missing values. Performance of all these Machine Learning algorithms greatly improved and they can give accurate results when the data is in full without missing values. So before applying these algorithms; dataset and its missing values are completely filled. To impute these missing values in the dataset there are numerous methods were proposed. In this paper we used micro array gene expression dataset; by introducing various percentages of missing values a new methodology is proposed to impute these missing values in the data set. The nature of micro array gene expression dataset is huge in dimensionality, so at first, we used recursive feature elimination method to select the best features which contributes much for model was selected then we apply the Ridge Regression for imputation. Imputations with other methods are compared. We evaluate the performance of all models by using the metrics like MSE, MAE, R-square. To select the best model in the set of models we used Normalized Criteria Distance (NCD) to rank the models under proposed metrics. The model with least NCD rank selected as the best model among other models, in our paper proposed model has got the lowest value among other models and considered to be the best model among other models.

Fatigue Damage estimation in vehicle from minimal instrumentation thanks to a mixed Engineering / Data Science approach

Monitoring damage and its potential causes in lab-tested structures requires extensive instrumentation that cannot be feasibly replicated in production assets. Since strain gauge instrumentation at a large scale is impractical, other proxy measurements for damage such as pressure, temperature or acceleration can be monitored and damage can be inferred from them. There is considerable difficulty in understanding not only which events lead to significant damage accumulation, but also in suitably and economically instrumenting assets in order to capture these data.We propose a general framework for exploring data which is inspired from the field of data science. Using a combination of engineering calculation packages and open-source data science tools, we show how engineers can further their understanding of the problem domain. We will present two case studies, each involving a different automotive component that is extensively instrumented with strain gauges, accelerometers and other sensors. We evaluate absolute damage from the strain data and, by using engineering indicators, data reduction, visualisations, correlations, we show how a minimal instrumentation subset can be identified for the purpose of damage approximation.

Review on Applications of R programming in Biological Data Analysis

The key programming language used in the field of data science is R programming, because of the freeavailability and ability to handle huge data. Packages developed in the R are used widely by scientists toobtain solutions for various real-world problems in several disciplines like healthcare, agriculture, andinformation technology. Next-generation sequencing data requires various bioinformatics analyses likequality checking, differential gene expression studies, and annotation which is facilitated by R packages andsoftware. Bioconductorprojectremains a hub of R packages that are open access and move the researchers tothe comfort zone for various kinds of analysis. In this review, we briefly discussed various packages usefulfor NGS data analysis and also explained how to use the packages for basic level analysis that benefitsresearchers having less exposure to R programming.

Data Science Education in <i>DS<sup>4</sup></i> Program at Osaka University

In the context of the human resource development program for independent data scientists led by Osaka University, this report provides a roadmap for human resource development in the field of data science by providing detailed information on the type of human resources to be developed, an outline of the main subjects, and administrative methods for a portion of the subjects.

Swarm Intelligence in Data Science: Challenges, Opportunities and Applications

The Swarm Intelligence (SI) algorithms have been useful in solving multifaceted optimization problems. SI Algorithms as the name suggests work on the simulation principle of behaviour of biological swarms. Today, there is lot of advancement in the field of Data Science; which comprises of large amount of Big Data that needs to managed as well as analysed. Traditional methods may only be applied to differentiable and continuous functions. For population-based approaches, Swarm Intelligence proves to be effective and efficient. In order to have better insight into applications of SI Algorithms in Data Science, and to accumulate a further reference for the future researchers in this field, this paper discusses scope of Swarm Intelligence in Data Science. On the basis of existing work; the following paper also focusses on the Challenges and Opportunities of SI in Data Science and also encourage to design more efficient algorithms in future that is capable to solve Data Science related problems in real world.