Skip to content

README.md

Latest commit

 

History

History
143 lines (99 loc) · 4.16 KB

README.md

File metadata and controls

143 lines (99 loc) · 4.16 KB
Raw

Data documentation

Included data

sample.csv contains eight fully synthetic learner trajectories used to exercise the self-regulated learning support policy.

No real learner, LMS, instructor, or institution is represented.

The dataset is intentionally designed around regulation states and support decisions, not hidden psychological inference.

SRL cycle represented

The synthetic schema separates several parts of a self-regulated learning cycle.

Forethought and planning

  • goal
  • goal_progress
  • next_action
  • success_evidence

These fields make planning explicit rather than assuming that low effort means poor planning.

Performance and strategy use

  • current_strategy
  • previous_strategy
  • strategy_effectiveness
  • strategy_changes
  • mastery
  • confidence
  • effort
  • stalled_minutes
  • recent_progress
  • monitoring_accuracy

The current code treats confidence as an explicit supplied learner-state signal. It does not infer motivation, self-efficacy, emotion, or mental state from behavior.

Learner control

  • learner_requested_help
  • support_preference
  • minutes_since_support
  • support_count

Supported preferences are:

  • normal
  • minimal
  • off

An explicit request for help can still produce support when the passive preference is off, because the learner is actively asking for assistance.

Reflection and adaptation

  • reflection_due
  • learner_response
  • next_strategy

The core package also contains a ReflectionRecord for learner-supplied strategy usefulness, perceived goal progress, confidence change, what worked, and a possible next strategy.

Deliberate synthetic trajectories

The data include:

  • L01: incomplete planning followed by plan formation and goal reflection
  • L02: productive struggle that should not be interrupted
  • L03: unproductive persistence, learner-requested help, and strategy change
  • L04: learner opt-out followed by an explicit request for help
  • L05: prompt cooldown and prompt-burden suppression
  • L06: monitoring uncertainty followed by improved monitoring evidence
  • L07: ineffective strategy followed by learner-selected strategy change
  • L08: minimal-support preference and end-of-goal reflection

These are test scenarios, not learner archetypes.

Important boundary

The repository does not estimate SRL state from raw clickstream data.

A real adapter would need to justify how observable events become state variables.

For example, strategy_effectiveness=0.2 should never appear from an undocumented black-box transformation.

Every derived state should retain:

  • source events
  • observation window
  • transformation rule/model
  • uncertainty or confidence
  • missing-data treatment
  • course/task context
  • timestamp/data-availability semantics

Productive struggle

The synthetic policy distinguishes productive and potentially unproductive struggle using supplied effort, recent progress, strategy effectiveness, and stalled time.

This is an operational policy definition, not a universal psychological classification.

A real study must validate whether these indicators identify useful moments for support.

Learner responses

The software supports explicit response categories such as:

  • accepted
  • declined
  • continued_without_help
  • requested_more_help
  • changed_strategy
  • completed_reflection
  • not_applicable

These responses are behavioral records of interaction with the support system. They should not be interpreted as motivation or ability.

Before real data are connected

Document:

  • target population and learning context
  • SRL framework used
  • task definition
  • goal representation
  • strategy vocabulary
  • state-estimation method
  • support policy version
  • threshold rationale
  • learner preference/opt-out mechanism
  • support cooldown
  • intervention history
  • prompt wording
  • learner response semantics
  • outcome measures
  • missingness
  • privacy/consent basis
  • retention and access controls

Do not commit

Do not commit identifiable learner traces, private messages, free-text reflections, grades, health/disability information, audio/video, restricted LMS exports, or proprietary course content.

Keep sensitive data outside Git and use an approved secure environment.