Educational Analytics: From Observation to Decision
How evidence about tasks, results, competencies and processes can inform learning and accountable university decisions.
An indicator is not an explanation. Educational analytics connects observable actions and outcomes with tasks, criteria, prior states and the conditions in which results were produced.
Specialised agents may identify patterns and prepare alternatives, but conclusions require scrutiny and an accountable person determines any action.
This is a proposed research architecture illustrated with examples, not a live analytics service.
Data
Recorded actions, results and process conditions
Stage 1 of 8
Data acquire meaning in context
Educational activity produces data about tasks, actions, results, resources and the conditions of collaboration. Meaning arises from links to aims, roles, criteria and prior states. An observation raises a question; interpretation offers possible explanations; an authorised participant selects an action and a way to verify its effects.
- 01
Record
Capture data about an event, action or outcome.
- 02
Connect
Relate the information to tasks, roles, criteria and processes.
- 03
Compare
Examine expected states and previous results.
- 04
Detect
Identify a significant change or restriction.
- 05
Explain
Develop possible causes and describe uncertainty.
- 06
Prepare options
Compare actions and their consequences.
- 07
Decide
An accountable participant chooses an authorised option.
- 08
Act
Carry out the approved change.
- 09
Review
Examine the resulting state and begin the next cycle.
Explore the University 5.0 analytical cycle
The separate research model illustrates how evidence, observations and responsible decisions connect to university processes.
Open model ↗Six levels of educational analytics
The model moves from data and relationships to observations, possible explanations, decision support and evidence of the effects of action.
Data sources
- Tasks
- Actions
- Results
- Competencies
- Schedules
- Resources
- Projects
- Decisions
- Events
- Feedback
How university activity becomes data
Each source must be connected to a particular task, action, responsible role and intended use.
Tasks
- Aim
- Deadline
- Criteria
- Status
- Dependencies
- Owners
Trace the provenance of an observation
A team has difficulty with argumentation
- 1
Project assignment
- 2
Argumentation criterion
- 3
Interim results
- 4
Expert comments
- 5
Repeated error
- 6
Analytical observation
Distinguishing data, observations and decisions
A measurement describes a property, an observation connects it with a context, an interpretation suggests a possible explanation, and a decision is made by an accountable participant.
- 01
Data
Three project stages finished later than planned.
- 02
Indicator
Completion times increased.
- 03
Observation
Delays follow the handoff between analytical and economic roles.
- 04
Interpretation
The team may be using inconsistent readiness criteria.
- 05
Decision
An accountable person approves a shared checkpoint and handoff format.
Investigate an analytical signal
Example: an interdisciplinary team's joint result becomes less coherent.
1. What was recorded
- Contradictions in the joint solution
- Repeated revisions
- Delayed integration
- Different quality criteria
A result must retain evidence of its origin
Results acquire educational significance when connected to the actions, participants, task, criteria and methods that produced them.
- 01
Initial task
Part of the traceable result record.
- 02
Participant role
Part of the traceable result record.
- 03
Actions
Part of the traceable result record.
- 04
Data and methods
Part of the traceable result record.
- 05
Result
Part of the traceable result record.
- 06
Quality criteria
Part of the traceable result record.
- 07
Feedback
Part of the traceable result record.
- 08
Related competencies
Part of the traceable result record.
- 09
Next action
Part of the traceable result record.
Five views of the same energy-saving task
Actions
- Gather data
- Clean data
- Build baseline model
- Compare scenarios
- Assess constraints
- Prepare reasoning
Competence as a system of linked evidence
A result can provide evidence for several competencies. A competence may require different tasks, contexts and expert assessments before it is confirmed.
- Actions performed
- Created result
- Criteria attained
- Application context
- Evidence currency
- Expert review
- Gaps requiring verification
- Transfer to new contexts
Evidence map: systems analysis
Energy-consumption analysis
What happens after an evidenced result
Next steps depend on verified abilities, aims, available roles and development gaps.
- 01
Persistent result
The participant consistently performs a class of tasks well. Suggested next step: Increase difficulty or broaden context.
- 02
Context gap
An ability is demonstrated in one field but not another. Suggested next step: Test transfer to a new context.
- 03
Method gap
An outcome exposes difficulty with a method. Suggested next step: Add explanation and guided practice.
- 04
Role readiness
A participant demonstrates coordination or integration. Suggested next step: Offer a new responsibility.
- 05
Project readiness
Competence is evidenced across educational settings. Suggested next step: Move towards applied project work.
Explore a next step
Illustrative profile: a data analyst with demonstrated scenario modelling and teamwork, seeking stronger economic reasoning and applied project experience.
Deepen analysis
A difficulty may belong to the system, not an individual
A persistent problem may follow from a task's architecture, weak handoffs, incompatible criteria or competition for resources.
- 01
Individual
- Task
- Action
- Result
- Pathway
- 02
Team
- Roles
- Dependencies
- Communication
- Shared outcome
- 03
Programme
- Modules
- Sequence
- Criteria
- Competence connections
- 04
Resources
- Time
- Spaces
- Equipment
- Expert availability
- 05
System
- Repeated constraints
- Interacting processes
- Persistent delays
- Rules
Investigate the source of recurring integration delays
Individual
A shared situation, different analytical questions
A student, educator, programme leader and university manager require different data, views and decision authority.
- 01
Student
- What is complete?
- What evidence exists?
- Where is the gap?
- What can I do next?
- 02
Lecturer
- How does the group perform?
- What difficulties repeat?
- Where do criteria differ?
- What needs expert review?
- 03
Programme leader
- Which modules create gaps?
- Where is workload concentrated?
- Which competencies are evidenced?
- Where are resource conflicts?
- 04
University management
- What restricts progress?
- How do programmes, projects and resources interact?
- What changes have systemic effects?
Student
Analytics must show the limits of what is known
Every inference depends on the source, context, method and plausible alternative explanations.
Evidence-quality criteria
- Currency
- Completeness
- Consistency
- Verifiability
- Fitness for purpose
- Identifiable source
- Sufficient context
- 01
Insufficient data
Further observations are required.
- 02
Contradictory sources
Evidence points to different states.
- 03
Alternative explanations
Several causes could fit the same event.
- 04
Changing context
New events can alter the interpretation.
- 05
Method limits
The analytical method only covers part of the process.
Qualitative states
- Sufficient evidence
- Verification needed
- Context changing
- Several explanations
- Limited applicability
The digital twin provides connected context
The proposed twin links participants, tasks, competencies, projects and resources. Observations and reviewed outcomes can update its state.
- 01
People
- Active tasks
- Roles
- Results
- Pathways
- Availability
- 02
Tasks
- Deadlines
- Dependencies
- Criteria
- Statuses
- Results
- 03
Competencies
- Evidence
- Contexts
- Currency
- Expert assessments
- 04
Resources
- Availability
- Workload
- Constraints
- Connected processes
- 05
Projects
- Stages
- Teams
- Results
- Risks
- External commitments
- 06
History
- Events
- Decisions
- Actions
- New states
- Consequences
AI agents support inquiry but do not replace responsible decisions
Specialists may structure evidence, identify patterns, question interpretations and prepare action scenarios. They must preserve sources, grounds and uncertainty for human review.
- 01
Analyst
- Connect evidence
- Identify changes
- Make observations
- Propose explanations
- 02
Critic
- Check reasoning
- Identify alternatives
- Compare claims with sources
- Flag weak grounds
- 03
Competence agent
- Map evidence
- Record application context
- Identify gaps
- Suggest tasks
- 04
Resource agent
- Assess workload
- Find conflicts
- Compare scenarios
- Describe consequences
- 05
Coordinator
- Link findings to processes
- Identify roles
- Prepare approvals
- Review effects
- 06
Governance agent
- Check authority
- Apply data rules
- Record grounds
- Specify permitted actions
How a recommendation is formed
A student team repeatedly revises its economic justification. The route gathers evidence, checks assumptions and sends options to accountable participants.
1. Sources
- Project results
- Expert feedback
- Revision history
- Assessment criteria
How analytics informs development of an educational programme
Several project modules produce strong discipline-specific outcomes but struggle to integrate them. The example examines the programme and its underlying processes rather than assigning unsupported blame to individual students.
1. Record data
Collect assignments, expert comments, revision histories, integration deadlines and role assignments.
Rules for trustworthy educational analysis
- 01
Defined purpose
Every analytical process addresses an explicit question or decision.
- 02
Traceable origin
Sources, relationships and grounds remain available for review.
- 03
Context
Interpretations include task, roles, criteria, time and constraints.
- 04
Uncertainty
Open questions, alternative explanations and limitations are included.
- 05
Accountability
Interpretation, approval and decisions belong to defined roles.
- 06
Limited access
Participants access only information relevant to their task and authority.
- 07
Effect review
Outcomes feed the next analytical cycle.
A traceable cycle of inquiry and responsible action
- 01
Connected evidence
Actions, results, competencies, processes and resources are interpreted together.
- 02
Provenance
Signals and conclusions remain linked to sources, criteria and responsible participants.
- 03
Distinct meaning
Data, indicators, observations, interpretations and decisions are not interchangeable.
- 04
Adaptive pathways
Next educational steps depend on verified results and goals.
- 05
System-level analysis
Recurring problems are examined across participant, team, programme and resource levels.
- 06
Role-sensitive views
Different participants need different analytical perspectives.
- 07
Expressed uncertainty
Quality, competing explanations and needed checks accompany recommendations.
- 08
Closed loop
Authorised action produces new evidence and a revised digital-twin state.
Educational analytics connects evidence, interpretation, authorisation, action and outcome review. It can inform the development of participants, educational programmes and University 5.0 processes.
About the model
This research concept is being developed by Sergey Shatokhin within IWXA, University 5.0 and EduOS Lad. It connects event-driven educational processes, competence evidence, the digital twin, specialist agents and accountable decisions. The examples describe a proposed architecture, not production measurements.