Synthetic learner insights

UCL-style demonstration • Autumn 2026 • assessment, engagement and risk patterns

Learners in view102Two synthetic modules
Mean course mark62.7%Recorded course total
Mean attendance81.6%Attendance indicator
Mean Moodle completion79.2%Activity completion indicator

Average assessment marks

Recorded marks only. Missing values are excluded from each assessment mean.

Module comparison

Mean course mark, attendance and Moodle completion

Engagement and attainment

Pearson correlation (r) with course total mark • full cohort only

Correlation measures association, not causation. Quiz and formative averages are performance measures, not merely activity counts.

Outcomes and risk flags

Full cohort counts. Risk is a derived flag, not an independent validation.

Completion versus course mark

Illustrative profiles from the source dataset. The original full-cohort Pearson correlation is r = 0.16. This plot shows selected documented outliers, not all 102 observations.

Select a point to inspect its Learner_ID.

Extensions, late submissions and feedback

Full-cohort mean course mark by status. These are associations.

Key observations

Assessment 1 has a lower mean than the two later assessments, but the sample sizes differ.
Moodle completion and attendance are weakly correlated with attainment in this synthetic cohort.
Quiz and formative scores show stronger correlations with total marks.
Extension and late-submission groups have lower average marks, without establishing cause.
Risk flags have exceptions and should never be used alone to judge learners.

Interpretation and practical actions

01Review assessment trajectories rather than a single mark.
02Combine multiple indicators before deciding what support to offer.
03Check missing and late submissions promptly and supportively.
04Review high-completion / low-attainment profiles separately.
05Audit risk rules and investigate exceptions with human judgement.
!Synthetic, observational and partly constructed data cannot establish causal effects.
Source: UCL_Mega_Synthetic_Learner_Assessment_Dataset.xlsx, Mega_Learner_Data. Aggregate dashboard: full cohort and module summaries; the selected-profile scatter is not a full-cohort plot. Self-contained HTML; share the downloaded file with colleagues. No external scripts, tracking or network requests.