This project evaluates a sample export of ~500 UK student accommodation listings across major university cities. The primary objective was to perform data quality auditing, clean corrupted records, analyze market pricing across property types, and formulate actionable recommendations for commercial stakeholders.
During initial exploratory analysis, several data integrity issues were identified and remediated before conducting commercial reporting:
| Issue Found | Action Taken | Business Justification |
|---|---|---|
Duplicate listing_id values |
Removed duplicate rows (kept single primary record) | listing_id must strictly act as a primary key. |
| City spelling inconsistencies | Standardized variants (e.g., Manchster → Manchester, Bristoll → Bristol, LONDON → London) | Prevents fragmented metric aggregation across cities. |
| Provider naming inconsistencies | Standardized casing and whitespace (e.g., unite students → Unite Students) | Ensures accurate count and review score evaluation per provider. |
| Text-formatted weekly prices | Converted strings to numeric float/integer types | Required for accurate mathematical calculations and statistical summaries. |
| Invalid weekly prices | Flagged extreme outliers (e.g., £0, negative rent, £6,800–£9,500/wk) outside the expected £110–£500 range | Extreme values represent data entry/ingestion errors; replaced with NULL to avoid skewing city averages. |
| Invalid review scores | Flagged scores outside the 1.0–5.0 range (e.g., 0.0, 6.2, 7.5) | Out-of-bounds metrics distort provider performance ratings. |
| Missing review scores | Retained missing values as NULL |
Avoided artificial imputation bias in the absence of ground-truth review sources. |
The top 5 accommodation operators by total listing volume across the dataset are:
- CRM Students (62 Listings)
- Downing Students (59 Listings)
- Collegiate (55 Listings)
- Unite Students (52 Listings)
- iQ Student Accommodation (50 Listings)
- Studio & 1-Bed Apartments consistently command the highest weekly rates across all UK markets (typically £280–£410/week).
- Non-En-Suite & Shared Apartments offer entry-level market rates (£140–£210/week).
- Data cleaning significantly reduced artificial variance caused by unvetted high-value anomalies (£6,000+/week).
Key Takeaway: Data quality defects directly threaten pricing accuracy and operator benchmarking.
The uncleaned dataset contained critical noise—including zero-pound listings, extreme price spikes, and out-of-bounds review scores. Directly importing unvalidated feeds into commercial reporting will distort average rental benchmarks and lead to flawed vendor evaluations.
- Automate Ingestion Pipelines: Implement automated validation rules at the ingestion phase (enforce unique
listing_id, range checks of £110–£500 for rent, and 1.0–5.0 validation bounds for review ratings). - Vendor Name Canonicalization: Establish a master reference dictionary for provider and city names to prevent report fragmentation.
- Flag & Review Workflows: Route out-of-bound pricing entries into an operational review queue before publishing to commercial dashboards.
listing_iduniquely identifies an accommodation record.- Weekly pricing reflects a standard single-occupancy contract duration.
- Rents outside the £110–£500 range represent processing anomalies based on target market parameters.
- Validation Boundaries: Are price limits (£110–£500) strict commercial hard-stops or soft thresholds subject to room tiering?
- Multi-Unit Listings: Do duplicate
listing_identries reflect multiple available room units or system ingestion duplicates? - Historical Inventory: Should unlisted/unavailable historical listings be soft-deleted or maintained in a separate historical archive?
- Clone this repository:
git clone [https://github.com/YOUR_USERNAME/uk-student-accommodation-analysis.git](https://github.com/YOUR_USERNAME/uk-student-accommodation-analysis.git)
- Open
data/processed/UK_Uni_Accom_Project_Cleaned.xlsxto review the clean dataset, dynamic pivot tables, and lookup mappings.