成果导向教育（OBE）：教学设计和教学实施目标是，学生通过教学过程最后所取得的学习成果。

管理数据：学校教务部掌握的诸如选课学生人数数据，专业及学科课程体系数据，以及学生各科课程成绩等数据等，这些也可用来作为混合教学的数据基础。此外，餐饮数据和社交数据也在管理数据范畴，餐饮数据涉及学生就餐的时间等。

宏观环境数据：这类数据包括产业数据和社会热点数据等，我们在教学过程、教学设计过程中也能用到。

学习过程的数据：是学生在学习过程中积累的大量数据。这类数据有线上与线下之分。线上的学习数据有线上登录、浏览、收藏、播放、暂停、快进、停止，有交互中的提问、回答、质疑、检讨，还有在测试环节中的答题时长、提交次数、正确率等；线下的课堂表现数据，包括通过摄像头、捕捉监测学生的情绪动作，分析他们的表情、动作，并基于此判断学生的情绪，再综合判断学生的学习态度、效果。

of “Rawlsian blindness” (Harvey, 2012, para. 12) where students’ demographic and prior educational records are not used from the outset to predict their chances of success and failure. On the other hand, is it ethical to ignore the predictive value of research evidence in particular contexts?

Higher education cannot afford to not use data

Transparency

tudent success is a complex and multidimensional phenomenon

Student identity and performance are temporal dynamic constructs

Students as agents

Learning analytics as moral practice

A learning analytics approach may make education both personal and relevant, and allow students to retain their own identities within the bigger system. Optimal use of student-generated data may result in institutions having an improved comprehension of the life-worlds and choices of students, allowing both institution and students to make better and informed choices, and respond faster to actionable and identified needs

learning analytics as the collection, analysis, use and appropriate dissemination of student-generated, actionable data with the purpose of creating appropriate cognitive, administrative and effective support for learners.

students and their learning behaviors, gathering data from course management and student information systems in order to improve student success

Ethical issues for learning analytics fall into the following broad, often overlapping categories:

he purpose of this survey was two fold: 1) to measure the use of learning analytics on higher education campuses , and 2) to measure awareness of and attitudes towards the use of learning analytics-designed tools at the course-level; specifically, to identify disparities and drive interventions to achieve academic equity.

equity within the context of higher education as access, opportunity, and advancement for all students to eliminate race and income as predictors of student success

l earning analytics is the intentional collection and analysis of student data within the l earning context in order to understand and optimize learning

The results indicated that outdegree centrality of a student, being in a group with higher average grade, communicating with the tutors, are the factors most correlated with better learning.

the data extracted for each post were the id of the sender, the id of the receiver, time the post was written, time it was modified, the content of the post, and forum and thread IDs. For each user we extracted user Id, role, PBL group and the posts he contributed. All participant IDs were removed to protect their confidentiality and

exploratory case study

The SNA mathematical analysis quantifies network parameters on individual actor levels, as well as group level. The mathematical analysis of SNA uses graph theory concepts to calculate metrics representing the nodes, the links or the network. Such as the distance to other actors in the network, the number of interactions with other actors, or how many times it bridged interac tions between communities.

SNA visualization renders relationships between actors in social networks by graphs known as sociograms;

can be implemented in two main ways, visualization, and mathematical analysis.

Social network analysis (SNA) is a collection of methods and tools that could be used to study the relationships, interactions and communications.

provide automated feedback on performance, inter vene during learning, provide scaffolding or support, UHFRPPHQGWXWRULQJSHUVRQDOL]HOHDUQLQJDQGVRRQ

7KHPRVWFRPPRQ1/3DSSURDFKWRDQDO\]LQJVWXGHQW ODQJXDJHLQ022&6KDVEHHQWKURXJKWRROVWKDWDQDO\]H HPRWLRQV

iSTART (Interactive Strategy Training for Active Reading and Thinking)

linear regression, dis FULPLQDQWIXQFWLRQFODVVLāHUV1D°YH%D\HVFODVVLāHUV support vector machines, logistic regression classi āHUVDQGGHFLVLRQWUHHFODVVLāHU

/DWHQW6HPDQWLF$QDO\VLV

identify n-grams

WKHSULPDU\ SXUSRVHRI1/3WRROVLVWKHDXWRPDWHGLQWHUSUHWDWLRQ of human language input.

Advisor level allows any member of staff, at Ulster, who has completed the training course to view a student's aggregated predictions of passing their course, in the current year, with a grade of 50% or higher. Data can be accessed by:   looking up an individual student viewing a specific module cohort viewing a particular programme cohort sorting by a number of variables (risk level, level of study, year of study, tariff point scores)

“随机森林”机器学习算法

The largest methodological difference between the two modelling approaches is in how they address the issue RIJHQHUDOL]DELOLW\,QH[SODQDWRU\PRGHOOLQJDOORIWKH data collected from a sample (e.g., students enrolled in a given course) is used to describe a population more generally (e.g., all students who could or might enroll in DJLYHQFRXUVH7KHLVVXHVUHODWHGWRJHQHUDOL]DELOLW\ are largely based on sampling techniques. Ensuring the sample represents the general population by reducing VHOHFWLRQELDVRIWHQWKURXJKUDQGRPRUVWUDWLāHGVDP  pling, and determining the amount of power needed to ensure an appropriate sample, through an analysis RISRSXODWLRQVL]HDQGOHYHOVRIHUURUWKHLQYHVWLJDWRU is willing to accept. In a predictive model, a hold out dataset is used to evaluate the suitability of a model IRUSUHGLFWLRQDQGWRSURWHFWDJDLQVWWKHRYHUāWWLQJ of models to data being used for training. There are several different strategies for producing hold out datasets, including k-fold cross validation, leave-one RXWFURVVYDOLGDWLRQUDQGRPL]HGVXEVDPSOLQJDQG DSSOLFDWLRQVSHFLāFVWUDWHJLHV

([SODQDWRU\ PRGHOOLQJLVDSRVWKRFDQGUHĂHFWLYHDFWLYLW\DLPHG at generating an understanding of a phenomenon. 3UHGLFWLYHPRGHOOLQJLVDQLQVLWXDFWLYLW\LQWHQGHGWR make systems responsive to changes in the underlying data.

Predictions are calculated in two steps. First, predictive models are constructed by machine learning methods from legacy data recorded in the previous presentation of the same course. Second, student performance is predicted by the predictive models and the student data of the current presentation.

OUA employs three machine learning methods: (1) Naïve Bayes classifer (NB), (2) Classifcation and regression tree (CART), (3) k-Nearest Neighbours (k-NN).

machine learning algorithms

OU analyse (OUA

Learning analytics dashboard

. TAM is more focused on the design and use of a technology whereas ARM on individuals’ attitudes to change.

Academic resistance model

(a) data-gathering of students’ activities in VLE, (b) data collection and data mining using learning analytics techniques, (c) visualisation of student activities in a widget, application, or VLE, and (d) refection by the teacher.

The “Random Forest” machine learning algorithm

The “Random Forest” machine learning algorithm

Multimodal learning analytic

Learning design: A missing piece in learning analytics

Discourse analytics: Understanding student communications

Social learning analytics: understanding student interactions through social network analysis

Predictive analytics: Supporting student learning by predicting future

l earning characteristic of the present-day l earning environment (Skrypnyk, Joksimovi ć , Kovanovi ć , Dawson, et al., 2015).

Pressey

Learning analytics draws on theories and methods from machine learning and data science, education, cognitive psychology, statistics, computer science, neuroscience, and social and learning sciences to name but a few

What Kinds of Pedagogical Uses Can Learning Analytics Serve and How Do They Support Learning?

social network analysis (SNA)

Topic modelling

Clustering, social network analysis, and topic modelling

Factor analysis

Correlation and association rule mining

Learning Analytics is the development and application of data science methods to the distinct characteristics, needs, and concerns of educational contexts and the data streams they generate for the purpose of better understanding and supporting learning processes and outcomes (see also an earlier definition by Siemens et al., 2011).

When prediction targets continuous variables (e.g. the delayed assessment score in Svihla et al., (2015), models such as linear regression, support vector machines and regression trees are commonly used. For categorical (including binary) outcome variables,

Learning analytics is not defined primarily by the source of data but by its size