Mining RFID Data: New Insights

Only a small percentage of companies have adopted RFID technology in their supply chain and service operations so far. However, the commitment of leading organizations (such as the US Department of Defense) and companies (such as Walmart, JC Penney and PG) is expected to eventually spread the use of RFID, just as the barcode technology has gained acceptance over time.

A schedule-based system is a system that operates on or contains within a schedule of  events and breaks at particular time intervals. A recent study by an international team of researchers provides the answer: A framework that systematically produces insights for schedule-based system, answers the following questions:

  • What set of actions and what type of data mining methods can be applied for analyzing RFID data?
  • How can the method be integrated so that we can obtain actionable insights regarding the system and domain?
  • How do these methods apply, given RFID data from a schedule-based system?

Since RFID (Radio Frequency Identification) systems are gaining increasing importance in industry, a schedule-based system with RFID has been illustrated in this study. These systems are extensively encountered in a variety of domains, ranging from manufacturing to social event management.

The developed framework is general and is applicable to any system of this kind. The applicability of the developed framework is illustrated through a case study, where real world data is analyzed using the introduced framework.

The research question to be answered in this paper is the following: Given RFID data from a schedule-based system in any domain (such as social event management, manufacturing, healthcare, etc.) what set of actions (including the data cleaning steps) and computations, and what type of data analysis and data mining methods can be applied, so that one can obtain actionable insights regarding the system and the domain?

The paper gives multiple contributions: an analysis framework, including its mathematical representation, for mining RFID data coming from a schedule-based system. The different types of insights that can be obtained through the introduced framework were enumerated and the corresponding algorithms that are needed in the analysis framework were presented. Finally, authors demonstrated the applicability of the developed framework through a case study.

RFID technology has a great potential for facilitating and enhancing the management of social events, where humans interact with each other over time and across different locations. Therefore, the authors have chosen to present the application of RFID in the context of a social event, specifically a scientific conference. Case study also describes how this data can be used in real-time for informing conference attendees and illustrates how the system operates.

While applicable to any schedule-based supply chain, production, or service operation, the data used in particular in the study belongs to the domain of social event management and comes from a four-day medical conference. Each attendee of the conference was provided with a unique RFID tag and their entry and exit times to the single conference hall were recorded. The total number of attendees was 272.

RFID Systems Data Mining

In the context of practical use, these kinds of systems give many options. For example, in social event management, ubiquitous information systems can use this information to suggest new people for professional social networks – when two attendees are identified in the conference, as entering and exiting similar events, the conference mobile application can recommend them each other to add into LinkedIn and other professional social networks. In the context of warehousing, an example scenario where the information on similar-behaving entities can be used is the following: Let us assume pallets of similar-behaving products entering a warehouse. Chances are high that these similar-behaving products will also exit the warehouse at around the same time. Therefore, the warehouse management system (WMS) software can be programmed so as to allocate neighboring locations for these two pallets. This way, these products can be put away and picked on the same route, saving time and cost.

The importance of RFID systems is ever increasing, and they find applications in a very wide range of domains. Authors consider there is a good foundation for further research, such as: extending the framework from the temporal domain to the spatiotemporal domain, by extending it to handle multiple locations. One of the important challenges in industrial applications is the challenge of big data. A possible future research can involve the development of the framework to accommodate for big data applications. The novelty of the research is the introduction of a data mining framework for the first time for this type of a system while the importance of the research lies in the fact it can be applied generally in a wide range of domains.

In-depth analysis with all the research steps of the study can be found in this link: https://ertekprojects.com/gurdal-ertek-publications/framework-mining-rfid-data-schedule-based-systems/

A New Data Mining Approach For Healthcare Center Operations

With constant growth of world population, demand for healthcare services is growing as well, increasing the healthcare costs by 2.4-7.5% per year until 2020 in various countries. Healthcare market in the United Arab Emirates (U.A.E.) is expected to grow 12.7% per year, until 2020 and will possibly grow even further due to development of medical tourism, an economic priority for U.A.E.

Analytical tools, namely, data mining and optimization can contribute to a large degree in achieving operational efficiencies.

Dr. Gurdal Ertek and Dr. Salam Abdallah from Abu Dhabi University, together with Dr. Mohsin Malik from The University of Melbourne, have developed a new data mining approach for healthcare operations. The approach has been applied using real-world data from one of the largest public hospitals in the U.A.E, demonstrating its applicability and benefits.

Among other key points, the study quantifies and analyzes the timing of patient arrivals in order to improve the patient admission process step through elimination of wasted time. It is crucial to point out the main performance measure in the study was lateness and the association of lateness with various factors.

The importance of this segment of the study is immense, as there are no academic studies on predicting lateness as a function of schedule day and time. There is, also, a significant gap of knowledge regarding the understanding of the lateness and the factors underlying lateness in healthcare centers. Dr. Ertek, Dr. Abdallah and Dr. Malik focused on the way patient arrivals into healthcare centers can be analyzed in order to come up with insights into lateness. They researched about factors associated with lateness and developed suggestions on how patient appointments can be scheduled to minimize this issue.

Study also presents an appointment scheduling approach, based on the fact that that lateness is dependent on the scheduled time, rather than being independent of it. One of the key contributions of the research study is a subsequent optimization model for scheduling appointments that may improve operational efficiency.

Some of the essential insights of this data mining approach are precious and include presentation of hidden patterns that generate information about patient arrival patterns, identification of independent variables that affect lateness, and determination of form and parameters of the best regression function. The three experts have concluded that the approach they developed can serve as an engine for determining optimal appointment day and times for healthcare centers.

Compared to other closely-related researches, this one is at least 10-fold larger, providing, so far, the most relevant and most detailed information regarding this subject. The study can be downloaded online and the approach can be applied by healthcare practitioners to analyze data from their own organizations.

Technology in the Service of Students: Rule-Based Expert Systems for Supporting University Students and Processes

Courses and grants are highly important aspects of higher education all over the globe. Nowadays, engaging with students is a more complex and student-demand driven process. Moreover, there is  an information overload on the student side, as considerable amount of information needs to be translated and clarified.

The counseling of students is usually performed by human advisors, bringing an immense administrative workload to faculty members, as well as other staff at universities. Bearing in mind that there are more than 15 million college understudies within the US alone, the significance of any help that can facilitate counseling and advising becomes obvious. A team of seven specialists from academia and industry (Gurdal Ertek, Gökhan Motor, Burak Aksoyer, Melike Avdagic, Damla Bozanlı, Umutcan Hanay and Deniz Maden) have developed two educational expert systems that can support and even replace human counsels regarding the matter of academic advising. The published research paper reports and discusses the development of the two educational expert systems applied at a private international university.

The first expert system is a course advising system, which recommends courses to undergraduate students, while the second expert system recommends grants to undergraduate students based on their qualification. Both systems have been implemented and tested using Oracle Policy Automation (OPA) software.

So, what are rule-based systems?

A rule-based system is a software, with the capacity to imitate the decision-making ability and expertise of human specialists. They are designed to solve problems as humans do, by using encoded human know-how or ability.  The know-how can be extracted and obtained directly through interaction with people, as well as from printed and electronic assets such as books, magazines, and websites.

There are many advantages of rule-based systems: they decrease costs since they reduce the requirement for human specialists; they are permanent; they can be utilized for different tasks and within a variety of information systems; they minimize errors usually made by humans. These aspects of rule-based systems increase functionality, reliability, and usability of such systems.

The first project presented in the study is Supporting Course Selection Decisions where the aim of authors was to create a rule-based system for Manufacturing Systems Engineering Program students at Sabanci University to prepare their schedules using Oracle Policy Automation (OPA).

The objective of the second one, Supporting Scholarship Decisions was to assist university students by giving suggestions on which scholarships in Turkey they are eligible for. There are numerous institutions that offer scholarships to university students , however, each foundation’s criteria and budget is distinctive. The aim of authors was to coordinate universities students with the scholarships that they are eligible for.  This can decrease a task that would take weeks to minutes.

Details of the research study can be found here:

https://ertekprojects.com/gurdal-ertek-publications/rule-based-expert-systems-for-supporting-university-students/

New Research on Wind Turbine Accidents Reveals Common Causes

According to 2016 statistics released by The Global Wind Energy Council (GWEC), the cumulative global wind energy capacity will double from 2016 to 2021, reaching 800GW by 2021.

While wind energy industry and the installation of wind turbines are growing, the drawbacks of wind energy are not always considered and evaluated. One particular problem with wind energy is wind turbine accidents. Wind turbine accidents include a multitude of ways in which wind turbines fail due to mechanical problems, nature, or humans.

A new study by a international research team reveals new insights into wind turbine accidents and failures. The study was conducted by Dr. Sobhan (Sean) Asian of La Trobe University, Melbourne, Australia, Dr. Gurdal Ertek, Abu Dhabi University, Abu Dhabi, UAE, and researchers from Singapore and Turkey. Researchers have analyzed the wind turbine accidents and failures in the four stages of a wind turbine’s life cycle, namely, transportation, construction, operation, and maintenance, and explored the relation between cause categories and the life cycle stages. Cause categories were identified as human, nature, and system & equipment.

Wind Turbine Life Cycle
Wind Turbine Life Cycle

The research revealed the association between deaths and injuries and various factors. The most important factor predicting the occurrence of death was discovered to be the stage in the wind turbine’s life cycle when the accident/failure took place. Other important factors include country, whether the wind turbine is onshore or offshore, and power of the turbine.

For occurrence of injury,  the most important predicting factor was discovered to be the power of the wind turbine. Other important factors included country, stage of life cycle when the event occurred, and accident year.

According to research results, accidents and failures caused by humans are most common during transportation. In construction and maintenance stages of the wind turbine’s life cycle, human causes are also the most common. System and equipment related accidents, on the other hand, take place in the form of electric system failures, material fatigue, and faulty material.  When accidents occur, the components most likely to be affected are blades and tower.

Dr. Gurdal Ertek, one of the researchers in the research team stated that: “This is the first study in the world where such a big collection of accident news was analyzed and the results were publicly shared. The data for our study is available now online in our research web site, so that other researchers can also download and analyze the data. We are hoping that this will be the first in a series of research we conduct and we are looking forward to contributing to the know-how on wind turbines and facilitating the safer use of this technology.”

Dr. Sobhan (Sean) Asian is Senior Lecturer at La Trobe University, Melbourne, Australia. Dr. Asian received his Ph.D. from School of Mechanical and Aerospace Engineering at Nanyang Technological University, Singapore. His areas of expertise include supply chain risk management, business analytics, and operations management.

Dr. Gurdal Ertek is an Associate Professor of Management at Abu Dhabi University, Abu Dhabi, UAE. Dr. Ertek received his Ph.D. from School of Industrial and Systems Engineering at Georgia Institute of Technology, Atlanta, GA, in 2001. His areas of expertise include data science, supply chain and warehouse logistics, and project management.