Pemodelan Faktor Risiko Keselamatan dan Kesehatan Kerja Berbasis Laporan Observasi Temuan Unsafe Condition dan Unsafe Action dengan Pendekatan Bayesian Network

Andini, Puti (2026) Pemodelan Faktor Risiko Keselamatan dan Kesehatan Kerja Berbasis Laporan Observasi Temuan Unsafe Condition dan Unsafe Action dengan Pendekatan Bayesian Network. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6010241103-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Salah satu upaya peningkatan budaya keselamatan kerja berupa sistem pelaporan ketidaksesuaian pada aspek Keselamatan dan Kesehatan Kerja (K3) menunjukkan bahwa seringkali data pada sistem ini belum dimanfaatkan dengan optimal dan belum dianalisis lebih lanjut. Dengan alasan tersebut, penelitian ini bertujuan untuk mengidentifikasi faktor risiko berdasarkan laporan temuan ketidaksesuaian K3 menggunakan pendekatan Text Mining dan membangun model network-based dengan pendekatan Bayesian Network. Penelitian ini menggunakan data dari sistem pelaporan perusahaan sebesar lebih dari 249 ribu data point selama periode tiga tahun dengan karakteristik data yang cukup variatif. Proses menstrukturkan data menggunakan metode text pre-processing, word embedding dengan S-BERT, dan clustering dengan Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN). Dalam penyusunan struktur model, penentuan arah panah hubungan antar nodes dihitung menggunakan Association Rule Mining. Model ini menunjukkan hubungan antar faktor risiko serta mengestimasi probabilitas peningkatan faktor risiko yang dipetakan. Penyusunan model mengikuti kerangka Human Factors Analysis and Classification Systems (HFACS) yang juga sebagai dasar dalam menyusun skenario pada pemodelan Bayesian Network dengan kelompok utama yaitu pengaruh organisasi (S1), pengawasan (S2), kondisi pemicu tindakan tidak aman (S3), dan perilaku kerja tidak aman (S4) dengan input data berupa data terstruktur hasil pemrosesan pada Text Mining. Analisis dilakukan berdasarkan kombinasi skenario berbasis kelompok utama pada kerangka HFACS yang dibandingkan dengan kondisi baseline (S0). Model berhasil mengestimasi probabilitas dengan rasio peningkatan risiko tertinggi berasal dari bentuk pengawasan. Hasil uji sensitivitas menunjukkan bahwa faktor pengawasan menjadi yang paling berpengaruh terhadap peningkatan Safety Risk Factor dengan probabilitas sebesar 0,982. Nilai mengindikasikan bahwa perubahan pada faktor pengawasan berkontribusi paling tinggi terhadap perubahan probabilitas faktor risiko. Arah pengendalian risiko K3 dapat dipetakan pada bentuk pengawasan terhadap kondisi aset maupun perilaku pekerja. Penelitian ini berhasil mengembangkan model analisis faktor risiko berbasis data secara kuantitatif serta dapat memetakan area yang perlu menjadi fokus perbaikan dalam pengendalian risiko K3.
===========================================================================================================================
An initiative to enhance workplace safety culture specifically through an Occupational Health and Safety (OHS) non-conformity reporting system reveals that the data within these systems is often underutilized and lacks further analysis. Consequently, this study aims to identify risk factors based on OHS unstructured reports using a text mining approach and to construct a network-based model employing Bayesian Networks. The study utilizes over 249.000 data points from a company's reporting system, spanning a three-year period and characterized by significant data diversity. Data structuring involved text pre-processing, word embedding via S-BERT, and clustering using Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN). During model construction, the directional relationships between nodes were determined using Association Rule Mining. The model illustrates the interrelationships among risk factors and estimates the probability of hazard occurrence associated with the mapped factors. The model structure adheres to the Human Factors Analysis and Classification System (HFACS) framework, categorizing elements into organizational influences, supervision, preconditions for unsafe acts, and unsafe acts. While utilizing structured data derived from the text mining process as input. Analysis was conducted based on scenario combinations derived from the HFACS framework's main categories. Modeling results demonstrate varying probability values across the baseline scenario (S0) and scenarios involving organizational influences (S1), supervision (S2), preconditions for unsafe acts (S3), and unsafe acts (S4), with the findings indicating an increase in Safety Risk Factors relative to the baseline condition. The model successfully estimated probabilities, identifying supervision as the source of the highest risk increase ratio. Sensitivity analysis revealed that supervision is the most sensitive factor regarding Safety Risk Factors, with a target probability of 0.982. The values indicate that changes in the supervision factor contribute most significantly to changes in the probability of risk factors. The direction of OHS risk control can be mapped based on the nature of supervision regarding asset conditions and worker behavior. This study successfully developed a quantitative, data-driven risk factor analysis model and identified areas requiring a focus on improvement for OHS risk control.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Organizational Influence, Supervision, Unsafe Conditions, Unsafe Behavior, Probability, Safety Risk Factors, Bayesian Network, Text Mining
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management H Social Sciences > HV Social pathology. Social and public welfare > HV551.5.I4 Hazard mitigation Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory. T Technology > T Technology (General) > T55 Industrial Safety
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26101-(S2) Master Thesis
Depositing User:	Puti Andini
Date Deposited:	04 Aug 2026 03:00
Last Modified:	04 Aug 2026 03:00
URI:	http://repository.its.ac.id/id/eprint/142157

Actions (login required)

View Item