Membangun Model Evaluasi untuk Pemasok yang Andal dalam Lingkungan Operasional Kompleks dengan Machine Learning Terstruktur (Studi Kasus : Rumah Sakit Pemerintah Provinsi Jawa Timur)

Fikri, Muhammad Naufal (2026) Membangun Model Evaluasi untuk Pemasok yang Andal dalam Lingkungan Operasional Kompleks dengan Machine Learning Terstruktur (Studi Kasus : Rumah Sakit Pemerintah Provinsi Jawa Timur). Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 6010241115-Master_Thesis.pdf] Text
6010241115-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (12MB) | Request a copy

Abstract

Pengadaan merupakan proses awal di setiap entitas rantai pasok. Nilai pengeluaran yang umumnya berkisar 60-70% untuk proses ini semakin menambah urgensitasnya. Pemilihan sumber pasokan yang tepat berguna dalam menghindari dampak buruk akibat disrupsi. Pendekatan evaluasi sumber berbasis portofolio yang mengkaraterisasi pemasok, akan menjadi solusi dalam memandu kondisi pasar yang kompleks dan semakin dinamis. Disisi lain, eksistensi disrupsi menuntut proses rantai pasok yang resilience, dengan melihat keandalan sebagai kriteria prospektif dalam menjamin operational excellence. Maka dari itu, penelitian ini diusulkan untuk berfokus melakukan evaluasi pemasok berdasarkan karakterisasi performa keandalannya, yang masih jarang menjadi tinjauan utama di beberapa studi terdahulu. Pendekatan baru diusulkan melalui pemanfaatan machine learning terstruktur. Principal Component Analysis (PCA) digunakan sebagai layer pertama dalam memahami data. Klasterisasi pemasok dilakukan oleh layer kedua menggunakan algoritma K-means. Algoritma long-short term memory (LSTM) menjadi layer terakhir dalam melakukan prediksi. Studi ini dilakukan dalam fasilitas kesehatan milik Provinsi Jawa Timur. Data yang berhasil dihimpun sekitar 3300 lebih, di eksklusi bertahap hingga didapat 3070 yang terdiri atas 16 pemasok unik. Ekstraksi fitur dari sumber data menghasilkan fitur-fitur dalam penyusunan 9 metrik performansi keandalan. Dua Principal Component (PC), yakni Service agility vs commerzialisation resilience, berhasil didefinisikan untuk menjelaskan informasi variansi kumulatif sebesar 74,2%. Pemetan ruang PC tersebut dimanfatkan untuk menyusun k-optimal sebanyak 3 klaster (nilai silhoutte score = 0,46), yang masing-masingnya membawa karakteristik unik. Hasil proyeksi dengan LSTM dilakukan dengan train yang menghasilkan validation loss sekira 2% dengan MAE 6,98 4,86% dan RSME 11,36 7,96%. Berdasar hasil proyeksi ini terdapat tendensi perubahan 0,51 0,23 satuan ruang PC dari posisi klaster awal. Beberapa pemasok diindikasi telah melenceng dari karakteristik klaster yang ada. Dengan demikian, wawasan ini memberikan gambaran bagaimana pemasok berkinerja di 33 titik historis dan 24 titik proyeksi kedepan.
===================================================================================================================================
Procurement constitutes a critical upstream process in supply chains, typically accounting for 60-70% of total organizational expenditure, thereby highlighting the importance of effective supplier selection. In increasingly complex and dynamic operational environments, portfolio-based supplier evaluation that emphasizes reliability is essential for supporting resilient supply chain performance and sustaining operational excellence. This study proposes a structured machine learning approach for supplier reliability evaluation, integrating Principal Component Analysis (PCA) for data representation, K-means clustering for supplier characterization, and Long Short-Term Memory (LSTM) networks for performance forecasting. The study is conducted within a public healthcare facility owned by the Province of East Java. Approximately 3,300 records were initially collected and subjected to a staged exclusion process, resulting in 3,070 valid observations representing 16 unique suppliers. Feature extraction from the data sources yielded nine reliability performance metrics. Two principal components (PC), namely service agility vs commercialization resilience, were identified explaining a cumulative variance of 74.2%. The resulting principal component space was utilized to determine an optimal clustering configuration of k = 3 (silhouette score = 0.46), with each cluster exhibiting distinct performance characteristics. LSTM-based projections were trained with a validation loss of approximately 2%, achieving a Mean Absolute Error (MAE) of 6.98 ± 4.86% and a Root Mean Square Error (RMSE) of 11.36 ± 7.96%. Based on these projections, an average positional shift of 0.51 ± 0.23 units in the principal component space was observed relative to the initial cluster positions. Several suppliers were identified as deviating from their original cluster characteristics. Consequently, these findings provide insights into supplier performance trajectories across 33 historical time points and 24 projected future periods.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Evaluasi Pemasok; Keandalan Rantai Pasok; Manajemen Pengadaan; Matriks Portofolio Pemasok; Seleksi Pemasok; Procurement; Supplier Evaluation; Supplier Portofolio Matrix; Supplier Selection; Supply Chain Reliability.
Subjects: H Social Sciences > HA Statistics > HA30.3 Time-series analysis
H Social Sciences > HA Statistics > HA31.35 Analysis of variance
H Social Sciences > HD Industries. Land use. Labor > HD30.23 Decision making. Business requirements analysis.
H Social Sciences > HD Industries. Land use. Labor > HD31 Management--Evaluation
H Social Sciences > HD Industries. Land use. Labor > HD38.5 Business logistics--Cost effectiveness. Supply chain management. ERP
H Social Sciences > HD Industries. Land use. Labor > HD39.5 Industrial procurement.
Q Science > QA Mathematics > QA278.55 Cluster analysis
Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics)
Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
R Medicine > RA Public aspects of medicine > RA960+ Hospitals
R Medicine > RA Public aspects of medicine > RA971 Health services administration.
T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T58.62 Decision support systems
T Technology > TA Engineering (General). Civil engineering (General) > TA169 Reliability (Engineering)
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26101-(S2) Master Thesis
Depositing User: Muhammad Naufal Fikri
Date Deposited: 29 Jan 2026 08:12
Last Modified: 29 Jan 2026 08:12
URI: http://repository.its.ac.id/id/eprint/131070

Actions (login required)

View Item View Item