Airbnb customer engagement
across London.

THE QUESTION

Turn marketplace data into decisions a marketing manager can act on.

The dashboard was designed around one main research question: which listing and host characteristics are associated with higher Airbnb customer engagement, and how can those insights support marketing strategy?

• Which London neighbourhoods receive the highest number of guest reviews?
• How does engagement vary across room types?
• Are there seasonal engagement patterns?
• How do pricing and portfolio structure relate to host engagement behaviour?

DATA PREPARATION

Python first, Tableau second.

I cleaned the dataset in Python before visualization: duplicate listing IDs were removed, records missing essential host or geographical fields were excluded, prices were converted to numeric values, missing prices were imputed by room type and neighbourhood, and implausible values were removed.

Tableau FIXED calculations were then used to create neighbourhood- and host-level measures so the analysis could compare market areas and portfolio structures rather than only individual listings.

VISUAL ANALYTICS

One dashboard, four complementary views.

• Choropleth map showing total review activity by London borough.
• Top 10 borough ranking for precise comparison.
• Average reviews-per-month by room type with a reference line.
• Multi-year seasonal engagement trends from January to December.
• Host-level scatterplot combining average price, reviews per month and portfolio size.
• K-means clustering to identify behavioural host segments.

SEGMENTATION

Host behaviour was analysed as a multidimensional problem.

The complex view used K-means clustering with Host Reviews Per Month, Host Average Price and Host Listing Count. The resulting segments included luxury-oriented lower-engagement hosts, mid-market professional hosts, budget/high-engagement profiles and a high-engagement outlier.

The analysis deliberately avoids claiming that price causes engagement; it treats the clusters as exploratory behavioural segments for marketing and marketplace analysis.

INTERACTIVITY

Geographic drill-down without breaking the global segmentation.

The borough map acts as a Tableau filter for the related descriptive views. Selecting a borough updates the associated charts, while the clustering view stays independent because it represents segmentation of the complete host population.

WHAT THIS DEMONSTRATES

Data cleaning, analytical modelling and visual storytelling in one workflow.

This project demonstrates my ability to turn a large marketplace dataset into a decision-oriented dashboard, combine descriptive analytics with unsupervised segmentation, and design coordinated visual views around a real stakeholder persona.