Mining Methodology Research Articles

An investigation of various machine learning techniques for mobile call data analysis for reducing call drop

Mobile call drop is a big concern for all telecommunications operators. In telecommunications, a cell phone call drop is a condition where calls have been cut off before the speakers have ended their conversation tone and before one of them has hung up. Therefore, the goal of this research is to design a predictive model that can estimate the mobile call drop data set of the mobile network. In order to address the disadvantage of simple statistical approaches, we have proposed and used data mining techniques, methods and methodologies in this study. In this paper, we propose a decision tree based on the mobile call drop data classification and prediction method. Experimental research is also being performed to test the proposed technique.

Multi-level process mining methodology for exploring disease-specific care processes

BackgroundPublic healthcare is a complex domain with many actors and highly variable protocols, which makes traditional process mining tools less effective and calls for specialized methods. AimThe objective of the work was to develop a generally applicable process mining methodology to explore care processes related to diseases. MethodsThe proposed methodology called Process Mining Methodology for Exploring Disease-specific Care Processes (MEDCP) is based on a systematic, step-wise refinement of the raw event logs by using such a multi-level expert taxonomy of events that encapsulates the professional concepts of the analysis. A treatment process is defined according to domain-specific rules to identify the starting (index) and closing events. Concepts from various levels of the taxonomy support the final process definition for an analysis that can deliver meaningful conclusions for domain experts. ResultsThe applicability of the methodology was demonstrated on two case studies in the cardiological and oncological care domains, in the public health care system in Hungary over a period of ten years. Thanks to the multi-level taxonomy, these studies successfully identified the most important high-level event sequence patterns and some key anomalies in the national care system, such as the significantly different behavior of low-volume vs. high volume care providers in the oncology study or the geographically connected, homogeneous clusters of providers with similar care spectra in the cardiology study. DiscussionAs the case studies showed, the proposed methodology can improve the efficiency of standard process mining methods, and deliver high level conclusions that are easy to interpret by domain experts. System-level insight into health care processes can serve as a basis for the optimisation and long-term planning of the whole care system.

Analyzing Madeira and Bermuda as Two Different Destinations: A Text Mining Approach

This study explores how tourists perceive two island destinations (Madeira and Bermuda) through reviews on social network platforms. An Artificial intelligence data mining methodology is used (text mining) to address the research problem. Authors collected TripAdvisor reviews, 1148 from Madeira and 635 from Bermuda (total pool of 1783 reviews). A database with three coding was generated, based on literature: Tourism Experience Model (TEM), Membership Level of reviewers (ML) and Categories. Data was analyzed through Meaning Cloud software. Results point out an overall customer satisfaction regarding both islands, but slightly superior for Madeira’s island. This research also discusses reviewers ranking level, the type of language in use, and the generation of negative content. It is possible to conclude that reviews can persuade others, namely their booking intention.

Automatic detection of maintenance requests: Comparison of Human Manual Annotation and Sentiment Analysis techniques

In the building management process, the collection of end-users' maintenance requests is a rich source of information to evaluate occupants' satisfaction and building systems. Computerized Maintenance Management Systems typically collect non-standardized data, difficult to be analyzed. Text mining methodologies can help to extract information from end-users' requests and support priority assignment of decisions. Sentiment Analysis can be applied at this aim, but complexities due to words/sentences orientations/polarities and domains/contexts can reduce its effectiveness. This study compares the ability of different Sentiment Analysis techniques and Human Manual Annotation, considered the gold standard, to automatically define a maintenance severity ranking. About 12,000 requests were collected for 34 months in 23 University buildings. Results show that current Sentiment Analysis techniques seem to limitedly recognize the role of technical words for severity assessment of requests, thus remarking the necessity of novel lexicons in the field of building facility management for automatic maintenance management procedures.

Examining the dimensions of big data privacy (Block chain solution for privacy protection)

Purpose: In the past, big data was the concern of governments and large commercial industries, and so these organizations had infrastructures separate from the public network to store and process their data. But today, big data is easily accessible to everyone through cloud infrastructure. In recent years, with the explosive development of the Internet, data storage and data processing technologies, privacy has been one of the major concerns in data mining Methodology: In this paper, after examining the dimensions and key components of flax data, privacy in big data; Blockchain solution for privacy protection is examined. Findings: This topic provides new opportunities for researchers in knowledge processing tasks. However, these opportunities often bring challenges as well. Originality/Value: Big Data means a very large database that can be structured or unstructured. Big data is a database that is constantly getting bigger with the increase of information. Also with the rapid advancement of the Internet, data storage and data processing technologies, big data privacy has attracted a lot of attention. Before applying big data technology to mass applications and applications, a basic issue called privacy must be explored.

Methodology for Exploring Patterns of Epigenetic Information in Cancer Cells Using Data Mining Technique

Epigenetic changes are a necessary characteristic of all cancer types. Tumor cells usually target genetic changes and epigenetic alterations as well. It is most beneficial to identify epigenetic similar features among cancer various types to be able to discover the appropriate treatments. The existence of epigenetic alteration profiles can aid in targeting this goal. In this paper, we propose a new technique applying data mining and clustering methodologies for cancer epigenetic changes analysis. The proposed technique aims to detect common patterns of epigenetic changes in various cancer types. We demonstrated the validation of the new technique by detecting epigenetic patterns across seven cancer types and by determining epigenetic similarities among various cancer types. The experimental results demonstrate that common epigenetic patterns do exist across these cancer types. Additionally, epigenetic gene analysis performed on the associated genes found a strong relationship with the development of various types of cancer and proved high risk across the studied cancer types. We utilized the frequent pattern data mining approach to represent cancer types compactly in the promoters for some epigenetic marks. Utilizing the built frequent pattern item set, the most frequent items are identified and yield the group of the bi-clusters of these patterns. Experimental results of the proposed method are shown to have a success rate of 88% in detecting cancer types according to specific epigenetic pattern.

Development of glaucoma predictive model and risk factors assessment based on supervised models

ObjectivesTo develop and to propose a machine learning model for predicting glaucoma and identifying its risk factors.MethodData analysis pipeline is designed for this study based on Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology. The main steps of the pipeline include data sampling, preprocessing, classification and evaluation and validation. Data sampling for providing the training dataset was performed with balanced sampling based on over-sampling and under-sampling methods. Data preprocessing steps were missing value imputation and normalization. For classification step, several machine learning models were designed for predicting glaucoma including Decision Trees (DTs), K-Nearest Neighbors (K-NN), Support Vector Machines (SVM), Random Forests (RFs), Extra Trees (ETs) and Bagging Ensemble methods. Moreover, in the classification step, a novel stacking ensemble model is designed and proposed using the superior classifiers.ResultsThe data were from Shahroud Eye Cohort Study including demographic and ophthalmology data for 5190 participants aged 40-64 living in Shahroud, northeast Iran. The main variables considered in this dataset were 67 demographics, ophthalmologic, optometric, perimetry, and biometry features for 4561 people, including 4474 non-glaucoma participants and 87 glaucoma patients. Experimental results show that DTs and RFs trained based on under-sampling of the training dataset have superior performance for predicting glaucoma than the compared single classifiers and bagging ensemble methods with the average accuracy of 87.61 and 88.87, the sensitivity of 73.80 and 72.35, specificity of 87.88 and 89.10 and area under the curve (AUC) of 91.04 and 94.53, respectively. The proposed stacking ensemble has an average accuracy of 83.56, a sensitivity of 82.21, a specificity of 81.32, and an AUC of 88.54.ConclusionsIn this study, a machine learning model is proposed and developed to predict glaucoma disease among persons aged 40-64. Top predictors in this study considered features for discriminating and predicting non-glaucoma persons from glaucoma patients include the number of the visual field detect on perimetry, vertical cup to disk ratio, white to white diameter, systolic blood pressure, pupil barycenter on Y coordinate, age, and axial length.

Data mining methodology for response to hypertension symptomology-application to COVID-19-related pharmacovigilance.

Potential therapy and confounding factors including typical co-administered medications, patient's disease states, disease prevalence, patient demographics, medical histories, and reasons for prescribing a drug often are incomplete, conflicting, missing, or uncharacterized in spontaneous adverse drug event (ADE) reporting systems. These missing or incomplete features can affect and limit the application of quantitative methods in pharmacovigilance for meta-analyses of data during randomized clinical trials. Data from patients with hypertension were retrieved and integrated from the FDA Adverse Event Reporting System; 134 antihypertensive drugs out of 1131 drugs were filtered and then evaluated using the empirical Bayes geometric mean (EBGM) of the posterior distribution to build ADE-drug profiles with an emphasis on the pulmonary ADEs. Afterward, the graphical least absolute shrinkage and selection operator (GLASSO) captured drug associations based on pulmonary ADEs by correcting hidden factors and confounder misclassification. Selected drugs were then compared using the Friedman test in drug classes and clusters obtained from GLASSO. Following multiple filtering stages to exclude insignificant and noise-driven reports, we found that drugs from antihypertensives agents, urologicals, and antithrombotic agents (macitentan, bosentan, epoprostenol, selexipag, sildenafil, tadalafil, and beraprost) form a similar class with a significantly higher incidence of pulmonary ADEs. Macitentan and bosentan were associated with 64% and 56% of pulmonary ADEs, respectively. Because these two medications are prescribed in diseases affecting pulmonary function and may be likely to emerge among the highest reported pulmonary ADEs, in fact, they serve to validate the methods utilized here. Conversely, doxazosin and rilmenidine were found to have the least pulmonary ADEs in selected drugs from hypertension patients. Nifedipine and candesartan were also found by signal detection methods to form a drug cluster, shown by several studies an effective combination of these drugs on lowering blood pressure and appeared an improved side effect profile in comparison with single-agent monotherapy. We consider pulmonary ADE profiles in multiple long-standing groups of therapeutics including antihypertensive agents, antithrombotic agents, beta-blocking agents, calcium channel blockers, or agents acting on the renin-angiotensin system, in patients with hypertension associated with high risk for coronavirus disease 2019 (COVID-19). We found that several individual drugs have significant differences between their drug classes and compared to other drug classes. For instance, macitentan and bosentan from endothelin receptor antagonists show major concern while doxazosin and rilmenidine exhibited the least pulmonary ADEs compared to the outcomes of other drugs. Using techniques in this study, we assessed and confirmed the hypothesis that drugs from the same drug class could have very different pulmonary ADE profiles affecting outcomes in acute respiratory illness. GJW and MJD accepted funding from BioNexus KC for funding on this project, but BioNexus KC had no direct role in this article.

Global Sustainability Reporting in the Automotive Industry via the eXtensible Business Reporting Language

Sustainability measurement has become one of the most important topics for automotive manufacturers. The financial reporting practice has been faced with an increased digitization and standardization process, which was enabled by the either voluntary or required implementation of the eXtensible Business Reporting Language (XBRL) platform. The obligatory adaption appeared in European legislation, which covers both financial, and non-financial information. In the current study, we have made a comparison of the most significant European and American automakers' sustainability reports data content to utilize the sustainability XBRL taxonomy adaptation. Using literature, the development of disclosure requirements was reviewed, following which the adaptation to global sustainability standards (GRI, SASB) in reporting was examined from a qualitative point of view, using a text mining methodology. It has been concluded that by the biggest automotive manufacturers the conditions are currently met to a limited extent, but the most significant obstacle is the lack of linking sustainability information to financial impacts. If the financial regulations require OEMs to adapt, and potentially disclose negative information (such as pollution data, penalties) may negatively affect investor perception. Based on the review XBRL is capable to become a global standard, however, reported contents should be carefully audited and linked to objectively verifiable financial data to provide relevant information to investors, decision-makers.

Seasonal and Multilevel Association Based Approach for Market Basket Analysis in Retail Supermarket

Market Basket Analysis is an observational data mining methodology to investigate the consumer buying behavior patterns in retail Supermarket. It analyzes customer baskets and explores the relationship among products that helps retailers to design store layouts, make various strategic plans and other merchandising decisions that have a big impact on retail marketing and sales. Frequent itemsets mining is the first step for market basket analysis. The association rules mining uncovers the relationship among products by looking what products the customers frequently purchase together. In retail marketing, the transactional database consists of many itemsets that are frequent only in a particular season however not taken into consideration as frequent in general. In some cases, association rules mining at lower data level with uniform support doesn't reflect any significant pattern however there is valuable information hiding behind it. To overcome those problems, we propose a methodology for mining seasonally frequent patterns and association rules with multilevel data environments. Our main contribution is to discover the hidden seasonal itemsets and extract the seasonal associations among products in additionally with the traditional strong regular rules in transactional database that shows the superiority for making season based merchandising decisions. The dataset has been generated from the transaction slips in large supermarket of Bangladesh that discover 442 more seasonal patterns as well as 1032 seasonal association rules in additionally with the regular rules for 0.1% minimum support and 50% minimum confidence.

Airbnb listings’ performance: determinants and predictive models

The present study analyzes Airbnb listings’ performance in terms of occupancy rate, number of bookings and revenue, by employing data mining methodologies. The research objective is twofold, to highlight the strongest determinants that influence customer’s purchase intentions and to propose reliable models capable of predicting the listings’ performance. The data set refers to the Airbnb market of Thessaloniki, Greece and contains explanatory variables about the hosts, lodgings, rules and quests’ ratings. Elaborated inducers derived from Artificial Intelligence are used as analytical tools. The interpretable models, sensitivity analysis and a proposed complex wrapper estimator provide evidence about the significance of specific explanatory variables and highlight the central role of the host. Random Forest outperforms its competitors and is proposed as the suitable classifier for the specific domain. The results and conclusions can be useful to individual hosts, professional listings’ managers, as well as legislative and taxation authorities.

Scenario-Mining for Level 4 Automated Vehicle Safety Assessment from Real Accident Situations in Urban Areas Using a Natural Language Process.

As the research and development activities of automated vehicles have been active in recent years, developing test scenarios and methods has become necessary to evaluate and ensure their safety. Based on the current context, this study developed an automated vehicle test scenario derivation methodology using traffic accident data and a natural language processing technique. The natural language processing technique-based test scenario mining methodology generated 16 functional test scenarios for urban arterials and 38 scenarios for intersections in urban areas. The proposed methodology was validated by determining the number of traffic accident records that can be explained by the resulting test scenarios. That is, the resulting test scenarios are valid and represent a matching rate between the test scenarios and the increased number of traffic accident records. The resulting functional scenarios generated by the proposed methodology account for 43.69% and 27.63% of the actual traffic accidents for urban arterial and intersection scenarios, respectively.

Prediction of Fatalities in Vehicle Collisions in Canada

Traffic collisions affect millions around the world and are the leading cause of death for children and young adults. Thus, Canada’s road safety plan is to reduce collision injuries and fatalities with a vision of making the safest roads in the world. We aim to predict fatalities of collisions on Canadian roads, and to discover causation of fatalities through exploratory data analysis and machine learning techniques. We analyse the vehicle collisions from Canada’s National Collision Database (1999–2017.) Through data mining methodologies, we investigate association rules and key contributing factors that lead to fatalities. Then, we propose two supervised learning classification models, Lasso Regression and XGBoost, to predict fatalities. Our analysis shows the deadliness of head-on collisions, especially in non-intersection areas with lacking traffic control systems. We also reveal that most collision fatalities occur in non-extreme weather and road conditions. Our prediction models show that the best classifier of fatalities is XGBoost with 83% accuracy. Its most important features are “collision configuration” and “used safety devices” elements, outnumbering attributes such as vehicle year, collision time, age, or sex of the individual. Our exploratory and predictive analysis reveal the importance of road design and traffic safety education.

Interpersonal Associations and Completing a Substance Use Day Treatment Program

ABSTRACT Purpose Social influence can be important to people’s substance use recovery. It functions in professional substance use treatment programs and in the 12-step approach to recovery. Method The study reported here employed clinical data mining methodology to study the role of interpersonal associations inside and outside of substance use treatment in predicting clients’ treatment success. Results In this study, the interpersonal associations available in treatment predicted greater odds of completing a day treatment program. Two interpersonal associations outside of treatment had an effect on successful program completion, but only one was in the hypothesized direction. Discussion Social learning theory and 12-Step philosophy offer a lens for viewing the role interpersonal associations might play in a client’s recovery process.

A case based, combined LCA and S-ROI methodology for sustainable mining in the Suceava County, Romania

LCA (Life Cycle Assessment) is a useful tool in decision-making for most businesses that opt for sustainability and offers the possibility to compare different products, processes, and value chain scenarios, both real and hypothetical. Coupled with S-ROI (Sustainable Return on Investment), the LCA has a great potential in using available data for existing mining sites in the North-East Region of Romania to assess the economic, social and environmental benefits of certain sustainability measures on a local and regional level. The article will explore this approach of combining the two methodologies: LCA (Life Cycle Assessment) and S-ROI (Sustainable Return on Investment), with necessary adjustments according to the characteristics of the local mining activities, to show key investment areas that can improve the value chain of copper exploitation and preparation in the mining perimeter Mănăila. The case of the copper mine in Mănăila offers a great opportunity to apply the current LICYMIN (Life Cycle of Mining) research and to use available Ecoinvent data for the copper ore by comparing the current value chain scenario with a proposed scenario that includes a different location for a mining ore preparation unit, closer to the quarry. The results will give an insight into the potential social and economic impact (the measure can translate into a higher local employment rate, better social stability, lower transportation costs, etc.) as well as the environmental impact (reduction of GHG emission, pollution, and energy efficiency) of the suggested changes.

A data mining-based cross-industry process for predicting major bleeding in mechanical circulatory support.

Over a third of patients, treated with mechanical circulatory support (MCS) for end-stage heart failure, experience major bleeding. Currently, the prediction of a major bleeding in the near future is difficult because of many contributing factors. Predictive analytics using data mining could help calculating the risk of bleeding; however, its application is generally reserved for experienced researchers on this subject. We propose an easily applicable data mining tool to predict major bleeding in MCS patients. All data of electronic health records of MCS patients in the University Medical Centre Utrecht were included. Based on the cross-industry standard process for data mining (CRISP-DM) methodology, an application named Auto-Crisp was developed. Auto-Crisp was used to evaluate the predictive models for a major bleeding in the next 3, 7, and 30 days after the first 30 days post-operatively following MCS implantation. The performance of the predictive models is investigated by the area under the curve (AUC) evaluation measure. In 25.6% of 273 patients, a total of 142 major bleedings occurred during a median follow-up period of 542 [interquartile range (IQR) 205-1044] days. The best predictive models assessed by Auto-Crisp had AUC values of 0.792, 0.788, and 0.776 for bleedings in the next 3, 7, and 30 days, respectively. The Auto-Crisp-based predictive model created in this study had an acceptable performance to predict major bleeding in MCS patients in the near future. However, further validation of the application is needed to evaluate Auto-Crisp in other research projects.

Predicting the flow zone indicator of carbonate reservoirs using NMR echo transforms, and routine open-hole log measurements: Insights from a field case study spanning extreme micro-structure properties

Predicting the flow zone indicator is essential for identifying the hydraulic flow units of hydrocarbon reservoirs. Delineation of hydraulic flow units is crucial for mapping petrophysical and rock mechanical properties. Precise prediction of the flow zone indicator (FZI) of carbonate rocks using well log measurements in un-cored intervals is still a daunting challenge for petrophysicists. This study presents a data mining methodology for predicting the rock FZI using NMR echo transforms, and conventional open-hole log measurements. The methodology is applied on a carbonate reservoir with extreme microstructure properties, from an oil “M” field characterized by a relatively high-permeability with a median of approximately 167 mD, and a maximum of 3480 mD. The reservoir from the M field features detritic, or vuggy structure, covering a wide range of rock fabrics varying from microcrystalline mudstones to coarse-grained grainstones. Porosity has a median of approximately 22%. Dimensional analysis and regression analysis are applied for the derivation of four transforms that appear to capture approximately 80% of the FZI variance. These four transforms are formulated using the geometric mean of the transverse NMR relaxation time (T2lm), the ratio of the free fluid index (FFI) to the bulk volume irreducible (BVI), the bulk density, the sonic compressional travel time, the true resistivity, the photo-electric absorption, and the effective porosity. Non-linear regression models have been developed for predicting the FZI using the derived transforms, for the carbonate reservoir from the M field. The average relative error for the estimated FZI values is approximately 52%. The same transforms are used as input for training a developed general regression neural network (GRNN), built for the purpose of predicting rock FZI. The constructed GRNN predicts FZI with a notable precision. The average absolute relative error on FZI for the training set is approximately 3.1%. The average absolute relative error on FZI for the blind testing set is approximately 22.0 %. The data mining approach presented in this study appears to suggest that (i) the relationship between the flow zone indicator and open-hole log attributes is highly non-linear, (ii) the FZI is highly affected by parameters that reflect rock texture, rock micro-structure geometry, and diagenetic alterations, and (iii) the derived transforms provide a means for further enhancement of the flow zone indicator prediction in carbonate reservoirs.

Analyzing Particularities of Sensor Datasets for Supporting Data Understanding and Preparation.

Data scientists spend much time with data cleaning tasks, and this is especially important when dealing with data gathered from sensors, as finding failures is not unusual (there is an abundance of research on anomaly detection in sensor data). This work analyzes several aspects of the data generated by different sensor types to understand particularities in the data, linking them with existing data mining methodologies. Using data from different sources, this work analyzes how the type of sensor used and its measurement units have an important impact in basic statistics such as variance and mean, because of the statistical distributions of the datasets. The work also analyzes the behavior of outliers, how to detect them, and how they affect the equivalence of sensors, as equivalence is used in many solutions for identifying anomalies. Based on the previous results, the article presents guidance on how to deal with data coming from sensors, in order to understand the characteristics of sensor datasets, and proposes a parallelized implementation. Finally, the article shows that the proposed decision-making processes work well with a new type of sensor and that parallelizing with several cores enables calculations to be executed up to four times faster.

A review of social media-based public opinion analyses: Challenges and recommendations

Compared with survey polls, social media can yield a better and more comprehensive understanding of public perceptions of special topics in a more scientific manner. However, despite this advantage, there seem to be limited investigations into the challenges in social media-based public opinion analysis. This study offers an understanding of the challenges in this field and some corresponding recommendations. Through a systematic literature review, we identify 54 papers to analyze and discuss issues related to data collection, data quality, and data mining. This paper summarizes a framework for social media-based public opinion analysis as well as the commonly employed data mining methodologies. We found that collecting public opinion data from Facebook and Weibo is difficult because of their restricted application programming interface and measures against Web Crawler. How to effectively and conveniently delete invalid data and how to design data mining methods for social media data, especially for those in Chinese, are still two main challenges in social media-based public opinion analysis. We claim that using multiple data sources, optimizing keyword settings, enhancing interdisciplinary cooperation, and paying more attention to the functional role of social media can benefit the development of social media-based public opinion analysis. This study also highlights the potential risks of releasing the personal information of the public in the use of social media data in research.

Making open educational resource videos on sustainable development: students’ attitudes, rationales, and approaches

ABSTRACT Aim and background The aim of this study was to investigate students’ attitudes, rationales, and approaches to making open educational resource (OER) videos (a form of OERs) on sustainable development (SD) in order to identify students’ competencies and effective pedagogical designs. Method Students registering for a teacher training course were invited to design and create pedagogies, make OER videos, and share the videos on YouTube on five SD topics: sustainable lifestyle, campus, community, enterprise, and earth development. The students provided their weekly journals and a final reflection on the whole process of making the OER videos on SD. This study used qualitative data analysis and text mining methodologies to analyse students’ process data of making OER videos on SD. Results and discussion The analysis results revealed that making OER videos on SD needed students’ ideational, inquiry, societal, and disciplinary competencies. Inferred pedagogical suggestions for practitioners to support students in making digital products on SD are to follow a linear pathway from ideational creation, inquiry process, societal transformation, to transdisciplinary reflection.