Geri Dön

A time aware behavioral clustering for retail customer segmentation: A methodological improvement over CLTV-based classification

Retail customer segmentation: A methodological improvement over CLTV-based classification

  1. Tez No: 1005533
  2. Yazar: MOMPOLOKI BANYANA TLHALEFANG
  3. Danışmanlar: DR. ÖĞR. ÜYESİ AHMET ŞENOL
  4. Tez Türü: Yüksek Lisans
  5. Konular: Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol, Computer Engineering and Computer Science and Control
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2025
  8. Dil: İngilizce
  9. Üniversite: Üsküdar Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Bilgisayar Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: Belirtilmemiş.

Özet

N/A

Özet (Çeviri)

This study proposes a time-aware and behaviorally driven methodology of customer segmentation through clustering and predictive modeling of customer behavior and value based on transaction data derived from the UK Online Retail II dataset (2010-2011). The dataset used 3,707 UK customers, with over 150,000 transactions occurring between January and June 2011. Ten behavioral features were generated from the transaction data, including Recency, Frequency, Monetary Value, basket size, customer tenure, and product assortment. The clustering analysis conducted K-Means clustering and validated its traits by using multiple clustering validation measures that included Silhouette Score (0.268), Davies-Bouldin Index (1.18), and Calinski-Harabasz Index (590.7). The K-Means clustering resulted in three significantly distinct clusters: 'Inactive', 'Moderate Value', and 'High Value Active'. Unlike the paper presented by Doodipala et al. (2024), which developed fully supervised and tested classification models that predicted CLTV categories based on complete datasets, this study tasked supervised models (Logistic Regression, Random Forest, AdaBoost, LightGBM, and XGBoost) with predicting membership in the behavioral clusters. The best-performing classifier was XGBoost, which had 99% accuracy, a macro F1-score of 0.88, and perfect recall (1.00) for the underrepresented high-value segment (7 customers). While this was a much smaller segment within the customer base than the educational purpose of using CLTV categories, this was a better outcome because it indicated an excellent random effort for characterizing value at 99% accuracy. CLTV was only used in the post-analysis validation phase to confirm the business relevance of the clusters. The results absolutely confirmed the cluster characteristics from the customer transaction data exhibited clear differences in value, confirming the behavioral clustering. Due to time component of the dataset, we could see how customer behavior and value were stable over 6 months despite customer activity dropping off after the first month. Additional more visualisations depicted customer behavior by presenting randomized visual tools that included a series of radar plots showing customer behaviors across clusters, cluster transition heatmaps, and Sankey diagrams. The work produced an interpretable segmentation framework that is realistic and more closely aligned to a business CRM customer segmentation than traditional RFM+ML techniques.

Benzer Tezler

  1. Tez No

    945902

    A frequency-based weighting mechanism for data embedding methodologies

    Veri gömme yöntemleri için sıklık tabanlı ağırlıklandırma mekanizması

    NAİL TAŞGETİREN

    Doktora

    İngilizce

    İngilizce

    2025

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolYıldız Teknik Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    PROF. DR. MEHMET SIDDIK AKTAŞ

  2. Tez No

    1020044

    Hava tahmin modelleri ve yüzdelik dilim regresyonu temelli hibrit güneş enerjisi öngörü sistemi geliştirilmesi ve uygulamaları

    Development and applications of a hybrid solar energy forcasting system integrating weather models and quantile regression

    AYŞEGÜL BİÇER

    Yüksek Lisans

    Türkçe

    Türkçe

    2026

    Meteorolojiİstanbul Teknik Üniversitesi

    Meteoroloji Mühendisliği Ana Bilim Dalı

    PROF. DR. AHMET DURAN ŞAHİN

  3. Tez No

    710603

    Prevention and detection of intrusions in wireless sensor networks

    Kablosuz algılayıcı ağlarda izinsiz girişlerin önlenmesi ve tespiti

    İSMAİL BÜTÜN

    Doktora

    İngilizce

    İngilizce

    2013

    Elektrik ve Elektronik MühendisliğiUnıversıty Of South Florıda

    Elektrik ve Elektronik Mühendisliği Ana Bilim Dalı

    PROF. DR. RAVI SANKAR

  4. Tez No

    978070

    Nöromorfik hesaplama ve memristor tabanlı ağ saldırı tespit sistemleri: Siber güvenlikte yeni yaklaşımlar

    Neuromorphic computing and memristor-based network attack detection systems: New approaches in cyber security

    YUSUF ETKA KÖYLÜ

    Yüksek Lisans

    Türkçe

    Türkçe

    2025

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Mekatronik Mühendisliği Ana Bilim Dalı

    PROF. DR. İLHAN KOCAARSLAN

  5. Tez No

    363909

    Enerji verimli kablosuz algılayıcı ağ tasarımı ve gerçekleştirilmesi

    Design and implementation of energy efficient wireless sensor network

    MEHMET ERKAN YÜKSEL

    Doktora

    Türkçe

    Türkçe

    2014

    Mühendislik Bilimleriİstanbul Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    PROF. DR. ABDÜL HALİM ZAİM

Geri Dön