Practical Approach of Time Series Data Clustering
Prof. Yukari Shirota (Professor, Gakushuin University, Japan)
There are many machine learning-based clustering methods for time series data. As shape-based clustering methods, k-Shape method and k-means with DTW (Dynamic Time Warping) distances are widely used currently. In clustering, a distance which is a measurement of similarity is very important. This keynote will provide the comparison among these distances visually, as well as Euclidean distances and a correlation coefficient-based distance. In stock price analyses, the period length of the data standardization becomes another important factor to obtain the desired result. The concrete example with different two lengths will be presented using real stock data.
In addition, when we compare the key trend of index movement such as a GDP growth rate, we would like to observe the data as it is without the standardization. Then because we cannot use k-Shape in the case, the amplitude-based clustering method becomes an appropriate method. Finally, when we write a paper with the clustering results, additional evaluation approaches might be required. For example, a comparison by SVD (Singular Value Decomposition) and a hierarchical clustering, or a chi-square test of the resultant clusters’ distribution would be effective for a proof of your hypothesis. The keynote will provide how to examine the clustering result to obtain what we want to prove, using the concrete example that we presented in one of the top conferences “ACM AI in Finance 2020” and so forth.
Prof Yukari SHIROTA (Professor of Gakushuin University) graduated from the Department of Information Science, Faculty of Science, the University of Tokyo, and then received a D.Sc. in computer science in 1998. As a researcher in the private sector, she conducted research for 13 years and then in 2001 she was involved in Faculty of Economics, Gakushuin University, Tokyo as Associate Professor. In 2002, she has become Professor, Faculty of Economics, Gakushuin University. In 2006 to 2007, she stayed at University of Oxford, Oxford, UK as an academic visitor. She is Fellow of Information Processing Society of Japan, a Board Member of the Japan Society of Business Mathematics, and a Board Member of the Japanese Operations Management and Strategy Association. Research fields are industry analysis by AI, visualization of data on the web, social media analysis, and visual education methods for business mathematics. She has read the paper in the top conference of the “AI in Finance” field: “An Analysis of Political Turmoil Effects on Stock Prices – a case study of US-China trade friction –“ (ACM AI in Finance 2020). She organized the special session titled “Awareness Technology for Economic and Social Data Analysis” in IEEE iCAST in 2019 and 2020, so that they can discuss the economics/social themes with the latest machine learning technologies.
For over 23 years, she has developed visual teaching materials for business mathematics and statistics, and for mathematics used in AI (see the following sites):
She has visited Indonesia for lectures and research, and published many papers with Indonesia researchers (see https://www-cc.gakushuin.ac.jp/~20010570/Indonesia/). Using tweets related to disasters such as COVID-19 and East Japan Great Earthquake, she has published topic extraction papers from the humanitarian standpoints with researchers of India, Indonesia, and Thailand.
How to Manage Your Research Footprints in the New Age of Scholarly Communication
Ms. Nobuko Miyairi (Scholarly Communications Consultant, Tokyo, Japan)
The origin of modern scholarly communication traces back to the establishment of national academies in the late 17th century. Personal correspondence between scientists was replaced by scholarly journals, in that peer review was embedded as a certification mechanism to ensure both authenticity and quality. As the research system expanded and diversified, each discipline developed its own style of communication and evaluation criteria, which were passed down from one generation to another through the academic apprenticeship.
It was only recently that the rapid and disruptive changes brought about by the advent of the Internet transformed the way we communicate, and scholarly communication was no exception. Scholars today communicate not only in traditional journals and books, but also through preprints, conference presentations, posters, social networks, and more. The Version of Record was and still is the ‘coin of the realm’ for many scholars, while data sets are gaining recognition as first-class research output. These changes are so acute and constant that the research community needs to adapt and update their practice continuously.This talk will provide an overview of new challenges emerging in the wholesale shift of scholarly content from print to online. In particular, scholarly metadata management is discussed in-depth to examine how vital it is for today’s scholars to consciously manage their research footprints so that they receive the credits that they deserve. This talk is aimed at both early career researchers and seasoned scholars who are open to exploring beyond their current research practice.
Nobuko Miyairi is a freelance consultant specializing in emerging aspects of scholarly communications. She is based in Tokyo, Japan, and provides strategic consulting for new business development in scholarly publishing and related services. Her service caters to a wide range of audiences both in and outside of Japan. She also teaches as a lecturer at the University of Tsukuba, advises the National Institute of Information and Communications Technology, and is a member of the DataCite APAC Expert Group.
Nobuko advised Paper Digest, an AI-based article summarisation service which received the Catalyst Grant by Digital Science in 2018; the People’s Choice Award at the Society for Scholarly Publishing (SSP) 41st Annual Meeting.Previously served as ORCID Regional Director, Asia Pacific, Nobuko is a strong advocate of persistent identifiers and open research infrastructures. She also held positions at Nature Publishing Group (now Springer Nature) and Thomson Reuters (now Clarivate), where she developed her business acumen and expanded her career interests. A librarian by training, Nobuko earned an MLIS from the University of Hawaii at Manoa.