Analisis Risiko Keterlambatan Pengiriman E-Commerce dengan Causal Discovery

Febrianti, Arnelita Putri (2026) Analisis Risiko Keterlambatan Pengiriman E-Commerce dengan Causal Discovery. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002221112-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (13MB) | Request a copy

Abstract

Keterlambatan pengiriman merupakan isu penting dalam layanan e-commerce karena dapat memengaruhi kepuasan pelanggan dan kinerja operasional. Penelitian ini bertujuan untuk menganalisis risiko keterlambatan pengiriman e-commerce menggunakan pendekatan causal discovery, dengan fokus khusus pada pengaruh kausal durasi persetujuan pembayaran terhadap keterlambatan pengiriman. Tiga metode causal discovery, yaitu Parallel Peter-Clark (Parallel-PC), NOTEARS-MLP, dan Model Averaging Hill Climbing (MAHC), digunakan untuk membangun struktur kausal antarvariabel. Graf yang dihasilkan kemudian digunakan untuk menentukan adjustment set dalam estimasi pengaruh kausal menggunakan Double Machine Learning (DML). Hasil penelitian menunjukkan bahwa variabel perlakuan, yaitu durasi persetujuan pembayaran, memiliki pengaruh positif dan signifikan secara statistik terhadap variabel luaran, yaitu keterlambatan pengiriman, dengan nilai Average Treatment Effect (ATE) sebesar 0.4875 untuk Parallel-PC, 0.5218 untuk NOTEARS-MLP, dan 0.4555 untuk MAHC. Hasil Conditional Average Treatment Effect (CATE) menunjukkan bahwa pengaruh variabel perlakuan terhadap variabel luaran bervariasi berdasarkan karakteristik pembayaran, waktu transaksi, wilayah pelanggan, dan karakteristik produk. Interpretasi menggunakan Causal-SHAP lebih lanjut menunjukkan bahwa variabel perlakuan merupakan fitur yang paling berpengaruh, diikuti oleh variabel terkait pembayaran, variabel wilayah, dan karakteristik produk. Temuan ini menunjukkan bahwa keterlambatan persetujuan pembayaran diperkirakan dapat meningkatkan risiko keterlambatan pengiriman e-commerce.
=============================================================================================================================
Delivery delay is a critical issue in e-commerce services because it can affect customer satisfaction and operational performance. This study aims to analyze e-commerce delivery delay risk using a causal discovery approach, focusing particularly on the causal effect of payment approval duration on delivery delay. Three causal discovery methods, namely Parallel Peter-Clark (Parallel-PC), NOTEARS-MLP, and Model Averaging Hill Climbing (MAHC), are employed to construct causal structures among variables. The resulting graphs are used to determine adjustment sets for causal effect estimation using Double Machine Learning. The results show that the treatment variable (payment approval duration) has a positive and statistically significant effect on the outcome variable (delivery delay), with Average Treatment Effect (ATE) values of 0.4875 for Parallel Peter-Clark, 0.5218 for NOTEARS-MLP, and 0.4555 for MAHC. The Conditional Average Treatment Effect (CATE) results indicate that the effect of the treatment variable on the outcome variable varies across payment characteristics, transaction timing, customer region, and product characteristics. The Causal-SHAP interpretation further shows that the treatment variable is the most influential feature, followed by payment-related variables, regional variables, and product characteristics. These findings suggest that delayed payment approval is estimated to increase the risk of e-commerce delivery delays.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Causal-SHAP, Causal Machine Learning, E-Commerce, Inferensi Kausal, Keterlambatan Pengiriman, Causal-SHAP, Causal Inference, Causal Machine Learning, Delivery Delay, E-Commerce
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA401 Mathematical models.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Arnelita Putri Febrianti
Date Deposited:	29 Jul 2026 01:02
Last Modified:	29 Jul 2026 01:02
URI:	http://repository.its.ac.id/id/eprint/139297

Actions (login required)

View Item