Student Outcomes Analytics Training Course

Participants can map the data required to answer a defined student outcomes question.

• SIS, LMS, assessment, attendance and advising data source mapping
• Grain and primary keys identified before joining student records
• Event timestamps reconciled across systems, time zones and academic terms
• Data lineage recorded from source extraction through published outcome metric
• Unique student identifiers matched using deterministic and probabilistic linkage rules
• Source-system limitations documented before interpreting missing or delayed records

Participants can apply privacy, ethical and governance controls to student analytics work.

• FERPA, GDPR and local policy obligations compared for student data
• Purpose limitation applied before collecting or combining sensitive student attributes
• Role-based access controls matched to analytical and intervention responsibilities
• Consent, notification and transparency requirements assessed for learner monitoring
• Data retention schedules aligned with academic, legal and operational requirements
• Ethics review questions for predictive risk scores and automated decisions

Participants can assess, clean and document a student-level analytical dataset.

• Completeness, validity, consistency, timeliness and uniqueness quality dimensions
• Profiling distributions, duplicates, invalid codes and unexpected category values
• Missingness mechanisms classified as MCAR, MAR and MNAR
• Outlier treatment based on domain rules, visual inspection and sensitivity tests
• Standardising programme codes, term labels and assessment grading scales
• Reproducible data-cleaning log documenting transformations, assumptions and exclusions

Participants can convert a student success issue into an analytical question and data specification.

• Problem statements framed around decisions, users, timing and intended action
• Unit-of-analysis selection for student, course, module and programme questions
• Target variables defined for withdrawal, failure, progression and completion
• Prediction windows separated from intervention windows and outcome windows
• Feature specification linking available signals to plausible student support actions
• Analysis brief documenting assumptions, exclusions, population and success criteria

Participants can identify meaningful outcome patterns using exploratory analysis and visualisation.

• Cohort analysis tracking student movement through enrolment, progression and completion
• Cross-tabulations exposing outcome differences by programme, campus and entry route
• Distribution plots revealing grade patterns, attendance variation and engagement spread
• Time-series views aligned with teaching weeks, assessments and registration deadlines
• Correlation analysis interpreted without treating association as causation
• Segmentation profiles combining academic history, engagement and support-service use

Participants can design dashboards that support timely student success decisions.

• Decision-centred dashboard requirements gathered from advisors, faculty and executives
• KPI hierarchy separating institutional outcomes, programme measures and student signals
• Cohort filters preserving valid comparisons across terms and academic levels
• Control charts distinguishing normal variation from meaningful performance shifts
• Accessible visual design using clear labels, colour contrast and annotations
• Dashboard governance covering refresh schedules, owners, access and escalation

Participants can apply appropriate statistical tests and interpret findings for student outcomes decisions.

• Confidence intervals used to communicate uncertainty around outcome estimates
• Chi-square tests comparing completion rates between defined student groups
• T-tests and ANOVA applied to assessment and engagement differences
• Effect sizes reported alongside p-values for practical interpretation
• Multiple comparisons managed when testing many programmes or demographic groups
• Regression coefficients interpreted with assumptions, controls and uncertainty stated

Participants can create useful and interpretable features from student activity data.

• Attendance rate features calculated against scheduled teaching opportunities
• LMS engagement measures derived from recency, frequency and submission behaviour
• Assessment trajectory features capturing improvement, decline and missed deadlines
• Academic history variables encoding prior credits, grades and repeated modules
• Support-service contact features interpreted cautiously to avoid stigma
• Feature documentation recording definitions, timing, source fields and known limitations

Participants can build an interpretable predictive model for a defined student outcome.

• CRISP-DM phases adapted for student outcomes analytics projects
• Training, validation and test samples separated by cohort and time
• Logistic regression modelled for interpretable withdrawal or failure probability
• Decision trees used to identify understandable student risk pathways
• Random forests compared with simpler models using stated performance criteria
• Class imbalance addressed through weighting, resampling and threshold selection

Participants can evaluate whether a student risk model is accurate enough for its intended use.

• Confusion matrices connecting predictions with intervention and non-intervention decisions
• Precision, recall and F1 scores selected against service capacity
• ROC and precision-recall curves compared for imbalanced student outcomes
• Calibration plots checking whether predicted probabilities match observed outcomes
• Thresholds set using false-positive costs, false-negative costs and capacity
• Temporal validation testing model performance on later academic cohorts

Participants can test a student risk model for unfair performance and harmful proxy effects.

• Protected characteristics distinguished from operationally relevant student support variables
• Proxy variables identified through correlation, context and feature review
• Group fairness metrics compared across demographic and programme populations
• False-positive and false-negative disparities examined before operational deployment
• Human review controls designed for high-impact intervention recommendations
• Model cards documenting intended use, limitations, performance and fairness findings

Participants can translate analytical findings into feasible student support interventions.

• Risk tiers linked to outreach channels, support offers and escalation paths
• Student support journeys mapped from signal detection through case closure
• Capacity planning matches advisor caseloads with predicted student demand
• Contact strategies tailored by timing, urgency, preference and support need
• Intervention protocols distinguish informational nudges from intensive case management
• Case-management records capture outreach, response, referral and resolution status

Participants can design an evaluation approach that tests whether interventions improve outcomes.

• Theory of change links outreach activities to expected student outcomes
• Comparison groups selected using randomisation, matching or quasi-experimental designs
• Difference-in-differences estimates intervention effects across cohorts and periods
• Propensity score matching reduces observed selection differences between groups
• Implementation fidelity tracked before attributing results to intervention design
• Cost-per-outcome analysis compares service resources with measurable student gains

Participants can plan controlled deployment and ongoing monitoring of student analytics products.

• Deployment checklist covering approvals, access, training and documented decision rules
• Model drift monitored through feature distributions, calibration and outcome changes
• Data pipeline controls detect failed refreshes, schema changes and late feeds
• Version control records datasets, features, model parameters and release dates
• Incident procedures address incorrect risk flags, data breaches and stakeholder concerns
• Periodic review determines retraining, retirement or redesign of analytical products

Participants can communicate student analytics findings clearly to technical and non-technical audiences.

• Executive narratives connect evidence, decisions, risks and recommended actions
• Faculty briefings explain patterns without ranking or stigmatising student groups
• Uncertainty statements clarify limitations, confidence intervals and model error
• Recommendation prioritisation uses expected impact, feasibility, cost and equity
• Data visualisations selected for advisory teams, programme boards and executives
• Challenge sessions document stakeholder concerns, alternative explanations and actions

Participants can build, review and present a complete student outcomes analytics product for a defined institutional decision.

• Capstone scope defines outcome, population, decision owner and intervention objective
• Data map identifies sources, lineage, quality risks and privacy controls
• Analytical dataset specification records features, targets, timing and exclusions
• Model and dashboard outputs support a defined student support workflow
• Evaluation plan sets comparison method, success measures and review timetable
• Peer review tests assumptions, fairness, feasibility and communication clarity