This project examines how Singapore's tourism sector recovered from the COVID-19 shock between 2019 and 2025, and whether the structure of visitor markets has returned to its pre-pandemic pattern or shifted to a new normal.
Using six interactive visualizations, the analysis tracks recovery dynamics over time, compares performance across source markets, and applies k-means clustering to identify distinct recovery archetypes. The project reveals structural transformation in Singapore's tourism composition, with strategic implications for policy and market prioritization.
DATA:
Monthly international visitor arrivals by place of residence are retrieved from Singapore Department of Statistics (SingStat).
Data processing was conducted in Python using Google Colab across multiple notebooks—one for each visualization—with common transformations including country name standardization, regional aggregation, and construction of a 2019=100 recovery index.
For geographic mapping, countries were matched to ISO3 codes. For clustering analysis, features including recovery performance, rebound speed, volatility, market share, and growth trajectory were standardized using z-score normalization and processed using scikit-learn's k-means algorithm.
All processed datasets are exported as tidy CSVs for visualization in Vega-Lite.
TOOLS:
Data processing used Python (pandas) in Google Colab with separate notebooks per visualization. Key challenges included standardizing country name variations and matching country names to ISO3 codes for geographic mapping.
Machine learning analysis used scikit-learn's k-means clustering; optimal k=3 was determined by comparing silhouette scores across k=2 to k=6. All interactive visualizations were implemented in Vega-Lite and embedded directly into the project HTML.
The Overall Recovery Path
International visitor arrivals to Singapore collapsed in early 2020 and remained near zero through much of 2021. Recovery began following border reopening in 2022, with indexed arrivals climbing from single digits back toward pre-pandemic benchmarks.
However, total visitor volumes have not consistently returned to 2019 levels even by 2025. The aggregate trend shows incomplete recovery with substantial volatility, but most importantly, this headline figure conceals significant variation across source markets pointing to structurally uneven recovery rather than uniform rebound.
The Geography of Recovery: Uneven and Asynchronous Reopening
Disaggregating by source country reveals asynchronous patterns. During early reopening in 2021Q4, only Bangladesh (31.6), Myanmar (11.7), and India (10.7) recorded meaningful recovery while most remained below index value 5.
This suggests recovery was driven by asynchronous border policies rather than global tourism conditions. Even by 2024-2025, convergence remained incomplete, with persistent disparities indicating structural disruption.
Winners and Losers: Tourism Market Recovery Relative to 2019
Comparing current performance to 2019 reveals fundamental bifurcation in market recovery. West Asia (117%) and Oceania (110%) surpassed pre-COVID levels, while most Asian and long-haul markets remain substantially below—Greater China at 87%, Southeast Asia at 82%, North Asia at 79%.
This divergence represents structural transformation rather than delayed recovery. Markets that drove Singapore's pre-pandemic tourism boom have not returned to form, while previously smaller markets have emerged as disproportionate contributors, setting the stage for strategic questions about future tourism priorities.
Singapore's tourism sector faces a portfolio challenge visible when plotting market size against recovery performance. Southeast Asia and Greater China—collectively 60% of 2019 arrivals—remain at just 82% and 87% of pre-pandemic levels. These are Singapore's largest source regions, and their underperformance creates substantial gaps in total volumes.
This translates to an aggregate shortfall of approximately 140,000 monthly arrivals below 2019 baselines, representing 11% of total pre-pandemic volumes.
Meanwhile, markets above 100% recovery—West Asia (116%), Oceania (106%), Americas (97%), Europe (95%)—contribute smaller baseline volumes. This creates fundamental tension: traditional high-volume markets are underperforming, while strong performers cannot fully compensate due to smaller size, raising critical questions about resource allocation and market prioritization strategies.
Differential recovery produces measurable shifts in visitor composition. Southeast Asia's share declined from 36% to 33%, while Oceania grew from 7% to 8% and West Asia expanded from under 1% to approximately 1%.
These shifts represent significant reallocation in a multi-billion dollar tourism economy, with implications for marketing spend, airline routes, and infrastructure investment. Singapore is not simply welcoming back 2019's tourist profile—the source market portfolio has been structurally rebalanced.
Recovery Archetypes
K-means clustering formalizes observed patterns into three data-driven market archetypes. Cluster 1 ("Slow Giant") consists of Greater China alone (23.9% market share), exhibiting slowest recovery (38 months to 50%) and highest volatility (38.8).
Cluster 2 ("High-Volume Laggards") comprises Southeast and South Asia (43.6% combined), showing faster initial recovery but lowest current performance (84.7%) with stable, flat growth. Cluster 3 ("Diverse Performers") encompasses six regions achieving highest recovery (96.3%), ranging from West Asia's 116% to North Asia's 78.9%.
The clustering algorithm achieves a silhouette score of 0.269 across k=3 clusters, with Greater China's 38-month recovery lag contrasting sharply with Diverse Performers' 26-28 month average, quantifying a 31% difference in recovery speed. This validates that Singapore's largest pre-pandemic markets (58.6% of 2019 arrivals) now constitute the weakest recovery segments.
Singapore's tourism recovery defies simple "return to normal" narratives. While aggregate arrivals have rebounded, underlying market structure has fundamentally shifted.
Traditional high-volume sources from Greater China (86.6% recovery) and Southeast Asia (83.0% recovery) collectively underperform by 14-17 percentage points, representing an aggregate monthly shortfall of approximately 195,000 arrivals below 2019 baselines.
Meanwhile, West Asia (116.3% recovery) and Oceania (106.4% recovery) exceeded pre-pandemic benchmarks by 10-16 percentage points, though their combined market share (7.8%) limits aggregate compensatory impact. K-means clustering (k=3, silhouette=0.269) validates this structural transformation through statistical segmentation.
The strategic question is whether resources should focus on recapturing lost share in familiar high-volume markets or pivoting toward emerging sources. The answer likely depends on whether current patterns reflect temporary policy lags or permanent shifts in travel preferences and regional connectivity.