Tokyo University of Science Case Study

Tokyo University of Science built a data analytics platform on Google Cloud to collect and utilize educational data on campus. By analyzing student learning and results with AI, the university is planning to build a foundation that will allow other universities to also utilize their data for delivering personalized education for every student.

Google Cloud results

Construct a data analytics platform that delivers results through data visualization, so results are easy to understand
Ensure that the platform can analyze data securely, cost-efficiently and in real-time, with Google Cloud offering scalable cloud storage
Reduce time spent in system operations so teams can focus on data analysis, while professors can tap on these insights to deliver more personalized lessons

Offering customized learning for students with a powerful data analytics platform

Education, like any other industries, should evolve alongside rapid cultural and social changes. This is the goal that Tokyo University of Science had in mind, with the university looking to accelerate and enhance student learning. It wanted to enhance its IT environment so it’s both flexible and student oriented, from improving the university's network to enabling students so they can learn effectively via computers.

That is why in recent years, Tokyo University of Science has been actively adopting forward-looking, emerging technology, such as creating an avatar of its president in the metaverse. As part of this initiative, the university is focusing on what it terms as "educational digital transformation (DX) to achieve a qualitative change in learning." One of its first steps is to reorganize the Educational Development Center into the Educational DX Promotion Center.

"The biggest mission of the Educational DX Promotion Center is to consider how education should change with the times with new educational approaches. As such, "educational program reform," "educational method development," and "educational environment improvement" are the three priority goals of the university's DX promotion plan," says Yasushi Idemoto, Vice President for Education, and Director of Organization for Education Advancement at Tokyo University of Science. "By specifically focusing on "educational method development" and "educational environment improvement," we are creating a learning support system based on enhancing the individual student's learning potential, with the help of artificial intelligence, as well as a web test to measure learning outcomes."

According to Yuki Watanabe, Center for Teacher Education, Organization for Education Advancement Professor at the Tokyo University of Science, the most important outcome in this initiative is to give back to the students. "The starting point is learning how to utilize the data the university has, so we can offer academic support to better enhance the individual student's learning. To do so, we had to learn how to collect, cleanse, and centrally manage the educational-related data that is scattered throughout the university, and return the results of these data analysis to each student. For this, we are tapping on Google Cloud."

Building a data analysis platform for real-time analysis

In charge of this construction is Daiwa Institute of Research, with the adoption of Google Cloud a proposal from the institute itself. "In this project, we are focused on setting up the system in a short period of time and at a low cost, so the use of public clouds was essential. In addition to the superior processing performance of BigQuery, when the university accumulates and analyzes various data such as class videos and audio in the future, the cost efficiency and scalability of cloud storage, along with unlimited capacity, are great benefits. What we like is the data analysis tool, Vertex AI, which is a very attractive tool when we eventually consider machine learning operations," says Michihiro Uchiyama, from Corporate System Department at the Daiwa Institute of Research.

"We are currently promoting a full cloud IT environment on campus, and we would like to make the data analysis platform particularly easy to integrate with other systems. Based on the aggregated data, we are adopting Google Cloud because we heard that it can perform advanced real-time analysis and information visualization."

—Hiroshi Matsuda, Manager of Public Relations Section, Management Planning Division, Tokyo University of Science

"Many faculty, staff and researchers at our university have been using Google Workspace and other Google products for some time, and I felt that we have a great affinity with Google Cloud," says Hiroshi Matsuda, Manager of Public Relations Section, Management Planning Division, at the Tokyo University of Science.

"We are currently promoting a full cloud IT environment on campus, and we would like to make the data analysis platform particularly easy to integrate with other systems. Based on the aggregated data, we are adopting Google Cloud because it can perform advanced real-time analysis and information visualization. In light of the current situation where even more real-time analysis is required in the future, we decided to use this data analysis platform. At the same time, we are also planning to aggregate the university's data."

Analyzing data with AI for personalized learning

For the new data analysis platform by Tokyo University of Science, when data is stored on Google Cloud from the existing on-premises environment through a closed network, Cloud Functions prepares the data and uses it as a data analytics platform. Security is ensured in the form of IP address restriction and identity access management (IAM) authentication for inbound data to Google Cloud, and restricted access for outbound data. The data is analyzed with Vertex AI, with Looker Studio used to deliver the analysis results via interactive dashboards and automated reports.

"Out of the features in Vertex AI, the AutoML function was especially useful. The ability to smoothly proceed with a series of tasks such as category estimation and notifying missing values makes the solution a good fit for us."

—Toshiya Sueyasu in the System Integration department, Daiwa Institute of Research

"Out of the features in Vertex AI, the AutoML function was especially useful. The ability to smoothly proceed with a series of tasks such as category estimation and notifying missing values makes the solution a good fit for us," says Toshiya Sueyasu in the System Integration department at Daiwa Institute of Research.

Watanabe says that through this initiative, he hopes to offer feedback to students and help guide them toward their desired learning outcomes. "The first thing we did with the platform was to analyze which learning behavior while in school contributed to high performance, based on the integrated academic data. In addition, we plan to build a machine learning model that predicts a student's grade point average at graduation, based on the integrated academic data in the same student's first year, in order to extract elements associated with high-performing graduates."

"We are now building the foundations for achieving our goals for the initiative. In order to provide the results of the analysis to each student, we need to examine the analysis results from a statistical perspective in the future. We hope that the AutoML function in Vertex AI, which allows us to look at the data directly and makes it easy to tune the machine learning model, will be useful."

—Yuki Watanabe, a Center for Teacher Education, Organization for Education Advancement Professor, Tokyo University of Science

At the same time, Tokyo University of Science has developed a system for online tests that leverages item response theory (IRT), a family of mathematical models used to measure a student's academic progress. This system is based on the system of Basic Mathematics Ability Research for High School Students, where results have accumulated over the years. By reviewing the conventional methods for collecting and measuring learning outcomes, the university intends to leverage this to deliver personalized guidance for each student.

"We are now building the foundations for achieving our goals for the initiative. In order to provide the results of the analysis to each student, we need to examine the analysis results from a statistical perspective in the future. We hope that the AutoML function in Vertex AI, which allows us to look at the data directly and makes it easy to tune the machine learning model, will be useful. With the support of Daiwa Institute of Research, we will further deepen our understanding of the function," says Watanabe.

"In the United States, one of the criteria for choosing a university is how advanced their digital transformation is. Naturally, this trend will also come to Japan, and I would like the university to pave the way for this," he adds.

Matsuda also points out that professors can spend more time on data analysis and better deliver lessons that are personalized for every student. "Professor Watanabe and other experts have been able to spend more time on analyzing data, which is a big achievement. We are now able to concentrate and work on more essential tasks."

Idemoto adds that the university will continue to make use of other tools like Google Cloud toward achieving an education that help every student reach their fullest potential. "Up to this point, I have focused on giving feedback to students, but there is also great potential for teachers to use data to understand each student better and choose appropriate teaching methods. I hope that through this initiative, we can extract factors that we were not able to anticipate before."

Tokyo University of Science: Delivering personalized learning through integrating academic results data

About Tokyo University of Science

Tell us your challenge. We're here to help.

About Daiwa Institute of Research

Google Cloud results

Building a data analysis platform for real-time analysis

Analyzing data with AI for personalized learning

Tell us your challenge. We're here to help.

About Tokyo University of Science

About Daiwa Institute of Research