London Airbnb Market Analysis

Spatial Data Science & Market Segmentation Strategy

Summary

This project utilises spatial data science and unsupervised machine learning (K-Means) to analyse the complex London short-term rental market. By processing real-world data from Inside Airbnb, the analysis identifies high-yield investment zones and segments the market into distinct operational personas, providing actionable insights for investors and policymakers.

Key Business Problems Solved

1. Investment Strategy: Identifying boroughs that offer the best balance between entry cost (price) and tourist demand (review volume).
2. Market Segmentation: Moving beyond simple pricing to understand the distinct business models operating in London (e.g., “Occasional Sharers” vs “Commercial Hotels”).

Methodology & Tech Stack

1. Data Cleaning & Engineering

• Raw Data Processing: Handled dirty data (currency symbols, mixed types) using Regex and Pandas.
• Outlier Removal: Filtered extreme luxury assets (£1,000+) to focus on the mass market.
• Standardisation: Applied StandardScaler to normalise features for clustering algorithms.

2. Geospatial Analysis

• Mapping: Utilised GeoPandas and contextily to project listing data onto official London Borough boundaries (EPSG:3857).

• Visualisation: Created professional choropleth maps with optimised labelling strategies for dense urban areas.

3. Machine Learning (Clustering)

• Algorithm: K-Means Clustering.

• Optimisation: Used the Elbow Method to determine the optimal cluster count ($K=5$).

• Interpretation: Decoded mathematical clusters into business personas (e.g., The Budget Kings, The Commercial Ops).

Key Insights & Visualisations

The Investment “Sweet Spot”

While Westminster is the most expensive, boroughs like Islington and Lambeth showed the highest efficiency—moderate pricing with exceptional review engagement.

Market Personas (The 5 Clusters)

The K-Means model identified five distinct operator types:

• Cluster 0 (The Occasional Sharers): Low availability (~88 days). Compliant with regulations.
• Cluster 1 (The Commercial Ops): High availability (~311 days). High regulatory risk.
• Cluster 2 (The Budget Kings): Low price (£103) but massive review volume. High turnover strategy.
• Cluster 4 (The Long-Term Lets): Extreme minimum nights (~274 days). Residential tenancies bypassing agents.