Back to ProjectsData Mining / ML
RESEARCH MONOGRAPH2025KOTA BEKASI, INDONESIA

Sukun Sales Segmentation

Spatial segmentation of potential sales areas using Hierarchical Agglomerative Clustering to identify regional sales characteristics and support data-driven marketing strategies.

PythonPandasNumPyScikit-learnSciPySpatial AnalysisData Visualization
Silhouette Score
0.6281
Davies-Bouldin Index
0.5347
Clusters
4
CCC Score
0.9292

Research Background & Problem Framing

Territory sales distribution for Sukun consumer goods often encounters uneven market penetration across different administrative sub-districts (kelurahan). Without quantitative spatial clustering, marketing resource allocation remains heuristic and sub-optimal.

This study applies Hierarchical Agglomerative Clustering (HAC) with Euclidean distance metrics to segment 17 kelurahan in Kota Bekasi, deriving distinct demographic and transaction volume profiles.

Dataset Specifications

Total Records:267 Transactions
Administrative Scope:17 Kelurahan
Geographic Focus:Kota Bekasi, ID
Study Period:2025

Methodology & Pipeline Architecture

01
Data Preprocessing
MinMax normalization, outlier screening, and transaction aggregation.
02
Distance Matrix
Pairwise Euclidean distance calculation across normalized dimensional vectors.
03
Agglomerative Linkage
Comparative evaluation of Average, Single, Complete, and Ward linkage variants.
04
Empirical Validation
Silhouette Coefficient, Davies-Bouldin Index (DBI), and Cophenetic Correlation (CCC).

Comparative Model Linkage Evaluation

(Scroll horizontally to inspect metrics →)
Linkage MethodSilhouette Score (Higher is Better)Davies-Bouldin Index (Lower is Better)Cophenetic Correlation (CCC)Verdict
Average Linkage (Selected)0.62810.53470.9292Optimal Quality
Complete Linkage0.58120.64200.8415Sub-optimal
Ward Linkage0.59400.61100.8650Moderate
Single Linkage0.43200.89200.7120Chaining Issue

Spatial Segmentation Visualizer

Kota Bekasi - Spatial Cluster Visualizer

17 Kelurahan across 4 Agglomerative Linkage Segments

GeoJSON Ready
Cluster 0: Very High Potential
Scope: 5 Administrative Kelurahan

High volume, rapid turnover sales territory with premium consumer density.

Sample Kelurahan:Pekayon JayaJaka SetiaMarga Jaya
GeoJSON Shapefile Integration: To render direct geospatial polygon boundaries with Leaflet tiles, place your GeoJSON in public/data/bekasi_kelurahan.geojson and enable the react-leaflet polygon renderer in this component.

Strategic Business Insights & Optimization

Cluster 0 Priority Allocation

Maintain high inventory buffer in Pekayon Jaya and Jaka Setia distribution hubs to prevent stockouts during peak retail cycles.

Cluster 2 & 3 Marketing Campaigns

Deploy targeted merchant incentives and introductory retail discounts to convert high-potential emerging territories into high-frequency zones.