The Application of Big Data and Learning Analytics in Universities Management: Practices from the Open University of China 魏顺平 博士 Shunping Wei May. 22nd, 2016

Outline 1.Big Data 2.Learning Analytics & Educational Data Mining 3.Application Cases

1.Big Data

The total amount of global data was 25ZB(2.5 trillion GB) in 2016. Big Data Era is Coming 25PB data is generated every day equal to the information amount of 1500 national libraries The total amount of global data was 25ZB(2.5 trillion GB) in 2016. As we know, the big era is coming. 25 PB data is generated everyday in the world, which equals the information amount of 1500 national libraries. The total amount of global data was 12ZB (Zettabyte) in 2016.?（同样是2016年的全球数据，对应幻灯片标题中的数字是25ZB？） Suppose the storage capacity of your hard disk is 1 TB(Trillionbyte), which can store about 1000 high definition movies. then you have to use 1.2 billion hard disks to storage the data generated in 2016. In education field, mass data is generated as digital education is developed swiftly. 1ZB=1024EB=1024*1024PB=109TB，100万块1TB硬盘 2020年全球数据将达到40ZB

OUC: Internet-based University

Six-Network Integration A learner development model centered on students and learning is established based on the "Six-Network Integration".

Big Data is Emerging Storing users’ activities data An onion model of educational big data Storing users’ activities data Activities Data Resource Data Status Data Basic Data Storing educational resources Storing operation status data of educational equipment and business Educational big data is emerging. In my opinion, educational big data is total amount, high volume, high velocity, and high variety data about educational factors and educational process. Here is an onion model of educational big data, which includes four layers. The inside layer is basic data, which stores national educational basic data such as students and teachers’ basic information. The second layer is status data which stores operation status data of educational equipment and business. The third layer is resource data which stores educational resources. The outer layer is activities data which stores users’ activities data. Storing national educational basic data

Big Data of OUC Basic data 13,000,000 records of learners’ basic information 8,900,000 records of graduates’ information 180,000,000 records of learners’ course selection Resource data 300+ Web courses 29,000 Micro-courses 510,000 minutes’ Video courses Digital library, more than 3,600,000 e-books, 11, 000 digital journals and magazines Behavior data More than 1 billion accumulative log data in recent 8 years

The Value of Big Data 1 2 3 4 5 6 For learning Personalized learning For instruction Targeted teaching For software Continuous optimization For research Learning rules discovering For management Fine management For desicion Policy making 从学习视角看，是如何利用大数据促进学生学业成功； 从教学视角看，是如何利用大数据提升教师教学效能； 从技术开发视角看，是如何利用大数据实现教育教学软件持续优化。 从研究视角看，是如何利用大数据发现教育和学习规律。 从管理视角看，是如何利用大数据优化配置教育教学资源。 从政策视角看， 是如何利用大数据聚焦于政策对象的微观层面，并获得迅速甚至达到实时与即时性的政策反馈，预测未来。

Database Administrators Big Data Centre in OUC BDC Data Scientists Data Analysts Database Administrators Software Engineers Director

2.Learning Analytics & Educational Data Mining

A Model of Learning Behavior action who what where When results If we describe learning activity, we could say a student assign a homework of one chapter in some time and gets ? . There exists five components in such a description as this figure shows. The action in the center, and who, do what, where, when, and get what results. This is a data model used to represent learning activity and learning activity is analyzed according to such a model.

Data Model for LA

Key Methods Many methods from data mining such as statistical analysis and visualization, relational mining, association rules mining, clustering are employed in this reaearch.

Typical Tools

Integrated LAS

Typical Examples

3.Application Cases

Data-driven Evaluation of Online Instruction and Learning Process Requirement description The headquarters of OUC evaluates online instruction and learning process of 44 branches each year. In the past, the headquarters employed many experts to evaluate the process by exploring some selected web courses of each branch artificially. The evaluation job would last 4 to 5 months. Application of big data and LA A data-driven Evaluation Index system for online tutoring and learning process of educational institution was constructed by using synthesis method of deduction and induction. Then the evaluation index system was used in evaluating the tutoring and learning process of 44 branches of the Open University of China based on about 110 million action logs. The practice showed that this evaluation index system could be used to present a high-level overview of online tutoring and learning of an educational institution, to present the differences among different institutions, and to diagnose the existing problems.

Width-depth-persistence Model for Learning Process Evaluation

Evaluation Dimension: Width

Evaluation dimension: depth

Evaluation Dimension: Persistence

Formula For Calculating Comprehensive Score Ybranch=ωTB(widthTB∧DepthTB∧PersistenceTB )+ωLB(widthLB∧DepthLB∧PersistenceLB )+ωI TL(widthITL∧DepthITL∧TimelinessITL) TB, Tutors’ Behavior LB, Learners’ Behavior ITL, Interaction between Tutors and Learners

Score in centesimal system Results a high-level overview of online tutoring and learning of an educational institution branches Original score Score in centesimal system BA 0.54 91 CD 0.43 86 JG 0.42 GX 0.41 85 LG 0.38 83 GU 0.34 79 GO 0.33 77 HE 0.32 76 JI TI 0.31 74 HU 0.29 70 SA 0.28 69

Results Present the differences among different institutions and diagnose the existing problems

Conclusions It’s helpful to use big data and LA appropriately in optimizing educational management, promoting the quality of education and instruction , and improving educational software design and development. It’s necessary and possible for open universities to use big data and LA in the teaching and learning process timely. It’s really valuable for gathering and storing data generated during the process of instruction, management and research.

One World, Multiple Models Learn from each other, develop together

Thanks wsp@ouchn.edu.cn