Annals of Data Science最新文献

Blockchain technology has ushered in significant technological disruptions within the operational management sphere, fostering value creation within operational management networks. In recent years, researchers have increasingly explored the potential applications of blockchain across diverse facets of operational management. Recognizing the pivotal role of comprehending prior research endeavors within any scientific domain for the development of a robust theoretical framework and a nuanced understanding of research progression in both the scientific realm and its practical applications, this study aims to identify areas where blockchain can be effectively employed. This objective is accomplished through an exhaustive systematic review of existing research on blockchain applications in the field of operations management. In pursuit of this goal, a comprehensive dataset comprising 9188 papers published up to the year 2020 is amassed and subjected to analysis employing life cycle analysis, bibliometrics, and textual analysis. The outcomes of this research elucidate the emergence of five distinctive clusters within the landscape of blockchain applications in operational management: Decentralized Finance, Traceability, Trust, Sustainability, and Information Sharing. These findings underscore the dynamic and evolving nature of blockchain’s impact in this domain.

This article introduces a new probability distribution suitable for modeling heavy-tailed and right-skewed data sets. The proposed distribution is derived from the continuous mixture of the scale parameter of the Pareto family with the Weibull distribution. Analytical expressions for various distributional properties and actuarial risk measures of the proposed model are derived. The applicability of the proposed model is assessed using two real-world insurance data sets, and its performance is compared with the existing class of heavy-tailed models. The proposed model is assumed for the response variable in parametric regression modeling to account for the heterogeneity of individual policyholders. The Expectation-Maximization (EM) Algorithm is included to expedite the process of finding maximum likelihood (ML) estimates for the parameters of the proposed models. Real-world data application demonstrates that the proposed distribution performs well compared to its competitor models. The regression model utilizing the mixed Pareto–Weibull response distribution, characterized by regression structures for both the mean and dispersion parameters, demonstrates superior performance when compared to the Pareto–Weibull regression model, where the dispersion parameter depends on covariates.

Corona virus (Covid-19) is a great danger for whole world. World health organization (WHO) considered it an epidemic. Data collection was based on the reports of World health organization for Covid-19 in Egypt. The problem of this study is to describe actual behavior of the virus using an appropriate statistical model. As WHO stated, Covid-19 behaves in the form of waves, therefore we thought that we should pay attention to seasonal and periodical models when identifying an appropriate model for this virus. The aim of this article is to introduce and study Periodical Autoregressive integrated Moving Average (PARIMA) models and compare them with the Autoregressive Integrated Moving Average (ARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) models to find optimal or approximately optimal model helps to predict the epidemiological behavior of the prevalence and so find reliable future forecasts of the number of Covid-19 injuries in Egypt. A numerical study using real data analysis is performed to establish an appropriate PARIMA model. The results supported the reliance of PAR (7) odel and its use for the purpose of forecasting. Extensive comparisons have been made between the estimated PARIMA model and some other advanced time series models. The forecasts obtained from the estimated PARIMA model were compared with the forecasts obtained from ARIMA (2, 2, 2) and SARIMA (1, 2, 1), (0, 0 ,1) models.

The purpose of this research is to explain the crisis management model of agricultural faculties in pandemic conditions. This descriptive-correlation research was conducted using a survey method. The staff and teachers of the agricultural faculties from universities in the Tehran province (493 people) constitute the statistical population of this research. Using Cochran's relationship and the size of the statistical population, the number of samples was estimated to be 240, and the samples were selected using the stratified random sampling method. The main tool of this research was a researcher-made questionnaire whose validity and reliability were tested and confirmed. In order to analyze data and test research hypotheses, structural equation modeling with a partial least squares approach and PLS Smart software were used. The results showed that formulation of laws and policies with a path coefficient of 0.137, diversification of financial resources with a path coefficient of 0.323, development and strengthening of infrastructure with a path coefficient of 0.245, communication with a path coefficient of 0.102, and human resources management with a path coefficient of 0.363 have significantly positive impacts on the pandemic crisis management, which directly explained 84.8% of the changes related to the variable of pandemic crisis management in the university. Moreover, pandemic crisis management at the university has a positive and significant effect on the sustainability of higher education with a path coefficient of 0.453, and it has the ability to predict 20.5% of changes in the sustainability of higher education.

A nonparametric test for the testing of scale parameters, is proposed in two-sample situation with random censored data. Random censored data are mostly encountered in clinical studies, where some individuals experience the event of interest (death); some are drop-outs or loss to follow-ups and some are still alive at the end of study. The performance of test is studied by comparing it with some existing tests in terms of asymptotic relative efficiency. Critical values required for the test are computed. Statistical power of the test is assessed through simulation study with varying sample sizes and varying censoring percentages. The working of test is illustrated by applying it to a real-life data set.

Due to the ever-increasing computing power and easy availability, social-networking platforms like Facebook, Twitter, etc. have become a popular medium to express one’s views instantly, be it about political situations, commercial products, or social occurrences. Twitter is a powerful source of information, whose data can be utilized to investigate the opinions of users through a process called Opinion Mining or Sentiment Analysis. Using the principles of Natural Language Processing and data science, this paper presents a comparative evaluation of multiple lexicon-based sentiment analysis algorithms to extract public opinion from tweets. The study explores the nuances of sentiment analysis using data science methodology, assessing how various lexicon-based algorithms may successfully identify and classify sentiments expressed in tweets from the Indian community about cryptocurrency.

In this paper, we introduce a new distribution for modeling count datasets with some unique characteristics, obtained by mixing the generalized Poisson distribution and the moment exponential distribution based on the framework of the Lagrangian probability distribution, so-called generalized Poisson moment exponential distribution (GPMED). It is shown that the Poisson-moment exponential and Poisson-Ailamujia distributions are special cases of the GPMED. Some important mathematical properties of the GPMED, including median, mode and non-central moment are also discussed through this paper. It is shown that the moment of the GPMED do not exist in some situations and have increasing, decreasing, and upside-down bathtub shaped hazard rates. The maximum likelihood method has been discussed for estimating its parameters. The likelihood ratio test is used to assess the effectiveness of the additional parameter included in the GPMED. The behaviour of these estimators is assessed using simulation study based on the inverse tranformation method. A zero-inflated version of the GPMED is also defined for the situation with an excessive number of zeros in the datasets. Applications of the GPMED and zero-inflated GPMED in various fields are presented and compared with some other existing distributions. In general, the GPMED or its zero-inflated version performs better than the other models, especially for the cases where the data are highly skewed or excessive number of zeros.

Wind energy is one of the fast evolving renewable energy sources that has seen widespread application. Therefore, research on its carrier, the wind turbine, is growing, and the majority of them concentrate on the diagnosis of wind turbine faults. In this paper, the vibration signals collected in the time domain by vibration monitoring were analyzed, and the fault characteristic parameters were identified. These parameters were then inputted into a genetic algorithm back-propagation neural network (GA-BPNN) for wind turbine fault diagnosis. It was found that the presence of defects in the wind turbine depended on the effective value, peak value, and kurtosis of the vibration signal. The overall recognition accuracy of the GA-BPNN was 94.89%, which was much higher than that of the support vector machine (88.7%) and random forest (88.35%). Therefore, it is feasible and highly accurate to extract fault characteristic parameters through vibration monitoring and input them into a GA-BPNN for wind turbine fault diagnosis.

The fresh engineering graduates are looking only for the popular jobs where the competition is high and the number of job openings is minimal, but they fail to look for the other job openings. The major problem is that the graduates fail to look at the number of requirements needed for a job role in the present and future. So there is a need for a prediction model that provides the number of job opportunities in a job role in the future. Many research studies have been carried out to predict the placement status of students, but they have not predicted the number of job opportunities in a job role. Many existing prediction models focus on improving prediction accuracy but fail to consider the handling of data fluctuations. When there is a data fluctuation, the predicted value deviates from the actual value. This paper presents a hybrid time-series prediction model called the enhanced integrated gated recurrent unit (EIGRU) Model to predict the number of job opportunities in a job role based on the company, salary, and experience. The proposed EIGRU model tries to minimize the divergence in the predicted value. The proposed time series prediction model is achieving a prediction accuracy of 98%. Based on the experimental evaluation of the Job dataset, the proposed model’s mean absolute percentage error and mean absolute error values are lower than the baseline models. As a result, the graduates will know about the number of job opportunities in their job role and make an effective decision.

Diabetic retinopathy is considered the leading cause of blindness in the population. High blood sugar levels can damage the tiny blood vessels in the retina at any time, leading to retinal detachment and sometimes glaucoma blindness. Treatment involves maintaining the current visual quality of the patient, as the disease is irreversible. Early diagnosis and timely treatment are crucial to minimizing the risk of vision loss. However, existing DR recognition strategies face numerous challenges, such as limited training datasets, high training loss, high-dimensional features, and high misclassification rates, which can significantly affect classification accuracies. In this paper, we propose a ResNet-50-based transfer learning method for classifying DR, which leverages the knowledge and expertise gained from training on a large dataset such as ImageNet. Our method involves preprocessing and segmenting the input images, which are then fed into ResNet-50 for extracting optimal features. We freeze a few layers of the pre-trained ResNet-50 and add Global Average Pooling to generate feature maps. The reduced feature maps are then classified to categorize the type of diabetic retinopathy. We evaluated the proposed method on 40 Real-time fundus images gathered from ICF Hospital together with the APTOS-2019 dataset and used various metrics to evaluate its performance. The experimentation results revealed that the proposed method achieved an accuracy of 99.82%, a sensitivity of 99%, a specificity of 96%, and an AUC score of 0.99 compared to existing DR recognition techniques. Overall, our ResNet-50-based transfer learning method presents a promising approach for DR classification and addresses the existing challenges of DR recognition strategies. It has the potential to aid in early DR diagnosis, leading to timely treatment and improved visual outcomes for patients.