Skills Matrix
Baseline at module start versus end-of-module position. Maps competencies and skills to a self-assessed level (Aware, Trained, Semi-Senior, Proficient, Expert) with supporting evidence for each rating.
The Professional Skills Matrix is updated twice during the module: at the start (baseline) and near the end (delta). Each row maps a specific skill to a self-assessed proficiency level and links to evidence on this site that supports the rating. Where the evidence is a Track 1 deliverable, the citation in parentheses resolves to the References page and the corresponding entry in the Evidence Index.
| Competency | Skill | Baseline (May 2026) | End-of-module (July 2026) | Evidence |
|---|---|---|---|---|
| Application | CRISP-DM operationalisation | Aware (cited in UAI essay) | Trained: designed and ran a five-script CRISP-DM-aligned pipeline with explicit phase ownership and feedback loops | Track 1 design document (Mella, 2026h); five-script pipeline (Mella, 2026l) |
| Application | Decision-register-driven methodology | Aware (UAI handover-style notes) | Trained: produced D-001 to D-018 register with alternatives, tier and citations per decision | Decision register (Mella, 2026g) |
| Application | Three-tier justification framework (T1 / T2 / T3) | None | Trained: designed and applied throughout Track 1 | Three-tier framework (Mella, 2026j) |
| Application | log-transform handling for right-skewed continuous targets | Aware | Trained: applied via log1p(price) regression target |
D-011 (Mella, 2026o); cleaning Rule 2.4 |
| Application | Categorical encoding (numeric mapping) for tree-free linear and clustering models | Aware | Trained: room_type_code, neighbourhood_group_code |
D-011 (Mella, 2026o); cleaning Rule 2.5 |
| Implementation | scikit-learn: regression (linear, multiple, polynomial) | Aware | Trained at module-aligned level (Units 3 to 4) | 04_regression.py (Mella, 2026l) |
| Implementation | scikit-learn: K-Means clustering with K-selection (elbow / SSE) | Aware | Trained at module-aligned level (Units 5 to 6) | 05_clustering.py (Mella, 2026l) |
| Implementation | pandas and NumPy: EDA and data cleaning at scale (about 50,000 rows) | Trained | Reinforced; designed an EDA / cleaning briefing pack pre-loaded with academic justification | Briefing pack (Mella, 2026i) |
| Methodology | Workstream split and handover-ready briefing pack design | None | Trained: produced a self-contained briefing pack consumed by another team member | Briefing pack (Mella, 2026i) |
| Methodology | Triple-source evaluation map (brief, tutor, rubric) | None | Trained: designed and applied | Triple-source map (Mella, 2026k) |
| Methodology | British English consistency, Harvard / Cite Them Right | Trained | Reinforced | This e-Portfolio; [Citation Style Guide internal] |
| Critical thinking | Iterative methodology with documented Phase 5 to Phase 1 loops | Aware | Trained: D-017 demand-mapping pivot as concrete instance | Demand-mapping pivot (Mella, 2026m) |
| Critical thinking | Defensive criticality (will-NOT-claim list) | Aware | Trained: explicit list inserted into the report | Will-NOT-claim list (Mella, 2026n) |
| Communication | Executive analytical reporting under tight word limits | Trained (IA e-Portfolio at 2,749 words) | Trained: different challenge - 1,000 words for executive Airbnb audience with full methodological rigour underneath | Track 1 report (in progress) |
| Team behaviour | Coordination and asynchronous-decisions architecture in distributed teams | Trained (IA Coordinator role) | Trained-plus: extended with the fault-tolerant team architecture frame (loose coupling per Weick, 1976) and applied to Group D’s reliability gradient | Unit 2 working agreement (Mella, 2026c); Team Project hub |
| Team behaviour | Equitable contribution documentation in support of fair assessment | Aware | Trained: maintained the contribution log throughout, with verbatim quotes and timestamps, in line with tutor’s 5 May guidance | Contribution log [internal]; Feedback page tutor commendation |
| Methodology | Pre-run design pattern: anticipating CRISP-DM Phase 4 to Phase 3 feedback before modelling starts | Aware | Trained: produced the EDA pre-run design document (Mella, 2026p) and cleaning script (Mella, 2026q) before the cleaning workstream began | EDA pre-run design (Mella, 2026p); cleaning script (Mella, 2026q) |
| Methodology | Risk-aware artefact design for cross-tool handovers | None | Trained: provided cleaning script as a separate .py file alongside the Word brief to neutralise Word’s smart-quote and indentation issues |
Cleaning script (Mella, 2026q) |
| Application | Literature-grounded data-cleaning rationale | Aware | Trained: each of seven cleaning rules anchored to a Tier-1 or Tier-2 citation (Bishop, Brownlee, Crawford, Harmadi, Patil, Tukey, whyalwaysme) | EDA pre-run design (Mella, 2026p) |
| Team behaviour | Partial-attendance decision-making with documented async-review rule | None | Trained: 9 May meeting (3 of 6 attended) ran the pre-agreed rule; decisions locked, summary email circulated within hours, async review opportunity preserved for absent members | 9 May meeting notes (Mella, 2026t) |
| Team behaviour | Workspace setup with role-appropriate access permissions | Aware | Trained: Google Drive workspace structured by workstream with differentiated permissions and explicit external-sharing policy | Drive workspace (Mella, 2026r) |
| Project management | Owner-named, dated project plan with input and output columns | Trained (IA Coordinator role) | Trained-plus: locked Track 1 plan through to 6 June 2026 with explicit dependency chain | Project plan and timeline (Mella, 2026s) |
| Communications | Structured written cadence for cross-functional handovers | Trained | Reinforced: three project emails (8 May, 9 May, 10 May) with named recipients, dated subject lines and explicit asynchronous-review invitations | 9 May meeting notes (Mella, 2026t); contribution log [internal] |
The matrix feeds the PDP, where each gap between baseline and end-of-module position becomes an actionable development goal.