MANHATTAN
RESIDENTIAL
CLUSTERING

Daegeun Kim
2025 Summer

Urban Analysis
Data Visualization
K-Mean Clustering

Manhattan Residential Buildings
Manhattan's residential buildings can be described by characteristics such as construction year, height, and transit access. These factors enable classifications into distinct clusters: socio-spatial, morpho-economic, accessibility-value, and beyond.

Residential Buildings:
  • Tax Class 1(Most residential property of up to three units)
  • All other property that is not in class 1 and is primarily residential

    • Residential Source Data

    The dataset was constructed by combining and processing information from multiple sources. All records were merged at the building level using the BIN (Building Identification Number) as the unique key. Since several original datasets were provided at the tax lot level, these entries were duplicated and expanded to align with individual buildings.

    K-means clustering was performed using selected datasets, with customized weightings applied to different variables to reflect their relative importance. The resulting clusters were then interpreted through qualitative analysis and assigned descriptive, context-specific names.

    ___________________________________________________________________

    Socio Spatial Clusters

    Weights:
    Building Height: 3
    Time to Closest Subway Station: 3
    Property Value Per Sqft: 1
    Elevator Access: 1
    Residential Area Share: 1

    This cluster integrates datasets on sociological traits, building form, and geographic location. The k-means analysis reveals that a significant share of Manhattan's residential stock falls into two main categories: Transit-Accessible Affordable Walk-Ups, concentrated around transit corridors, and Peripheral Walk-Ups, located at the city's edges with more limited accessibility.
    K-means clustering was performed using selected datasets, with customized weightings applied to different variables to reflect their relative importance. The resulting clusters were then interpreted through qualitative analysis and assigned descriptive, context-specific names.

    ___________________________________________________________________

    Architectural Clusters

    Weights:
    Building Stories: 3
    Building Height: 3
    Elevator Access: 1
    Residential Gross Area: 2
    Residential Area Share: 2

    The clusters are dominated by small-scale walk-ups. These buildings are relatively homogeneous, with low height, modest gross floor area (GFA), and high residential area share, while the other types in this cluster exhibit much wider variation across height, GFA, and residential share. Walk-ups also form an absolute majority, outnumbering all other types combined.
    K-means clustering was performed using selected datasets, with customized weightings applied to different variables to reflect their relative importance. The resulting clusters were then interpreted through qualitative analysis and assigned descriptive, context-specific names.

    ___________________________________________________________________

    Evolutionary Clusters

    Weights:
    Building Construction Year: 4
    Building Height: 2
    Property Value Per Sqft: 1
    Building Class: 1

    This cluster draws on datasets related to evolutionary traits, with a focus on temporal factors such as construction year alongside complementary attributes that influence the development of residential buildings. The k-means results indicate that a large portion of the building stock belongs to Early 20th-Century Multi-Family Low-Rises, followed by a secondary concentration of Mid-20th-Century Mid-Rises, reflecting the historical layering of Manhattan's built form.
    K-means clustering was performed using selected datasets, with customized weightings applied to different variables to reflect their relative importance. The resulting clusters were then interpreted through qualitative analysis and assigned descriptive, context-specific names.

    ___________________________________________________________________

    Economic Financial Clusters

    Weights:
    Building Height: 3
    Time to Closest Subway Station: 3
    Property Value Per Sqft: 1
    Elevator Access: 1
    Residential Area Share: 1

    This cluster integrates datasets capturing economic and financial dimensions, primarily those tied to property valuation and related indicators. The k-means analysis shows that many buildings fall within the category of Affordable Walk-Ups located near transit corridors, while a distinct subset displays exceptionally high property value growth rates, highlighting stark contrasts in financial performance across the housing stock.