Primantara, Ari (2021) Analisis Faktor – Faktor Yang Mempengaruhi Performa Transportir Darat Dengan Model SCOR dan Supervised Machine Learning Untuk Meningkatkan Performa Supply Chain di Industri Pupuk. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
09211950095004-Master_Thesis-Ari Primantara.pdf - Accepted Version Download (3MB) | Preview |
Abstract
Perkembangan persaingan bisnis saat ini membuat semakin pentingnya peranan pengelolaan supply chain untuk menunjang bisnis perusahaan, termasuk dalam industri pupuk. PT Petrokimia Gresik sebagai perusahaan yang bergerak di industri pupuk, memiliki target penyaluran pupuk terbesar pada area jawa bali dengan distribusi produk menggunakan armada transportir darat. Untuk menjamin ketercapaian target perusahaan, maka diperlukan adanya performa supply chain yang baik. Penelitian ini bertujuan untuk meningkatkan performa supply chain industri pupuk dengan pendekatan baru, yaitu mengkombinasikan model pengukuran performa supply chain menggunakan SCOR Model dengan Predictive Analytics menggunakan Supervised Machine Learning dengan algoritma random forest (RF), support vector machine (SVM) dan artificial neural network (ANN).
Penelitian ini telah meliputi tahapan descriptive dan predictive dan dapat dikembangkan pada penelitian selanjutnya dengan melanjutkan pada tahapan prescriptive melalui optimalisasi usulan perbaikan, peningkatan periode analisis selama satu tahun dengan data realtime. Dari pengukuran performa supply chain didapatkan atribut supply chain responsiveness adalah atribut performa kritis dengan parameter KPI perusahaan yang tidak tercapai adalah “ketepatan waktu pengambilan order” dengan ketercapaian hanya 37,76%. Kemudian dilakukan analisa variabel yang berpengaruh dengan feature selection , didapatkan bahwa terdapat 9 variabel yang berpengaruh terhadap ketepatan waktu pengambilan. Sembilan variabel berpengaruh ini dilakukan analisis deskriptif dan analisis predictive klasifikasi machine learning dengan algoritma Random Forest (RF), Support Vector Machine dan Artificial Neural Network didapatkan bahwa algoritma RF memiliki tingkat prediksi yang paling baik dengan akurasi sebesar 83,91% dan nilai area under curve (AUC) sebesar 0,88 (good classification). Dari gini ratio pada algoritma RF, didapatkan bahwa terdapat 4 faktor yang sangat signifikan terhadap ketepatan waktu pengambilan adalah Qty order pengiriman (POSTO), jenis produk, tanggal order POSTO dan biaya kirim. Rekomendasi perbaikan dilakukan dengan membuat model prediksi keterlambatan serta usulan alur baru order POSTO, berdampak pada estimasi peningkatan performa supply chain sebesar 136% dan estimasi penghematan finansial sebesar Rp 4,37 Milyar
====================================================================================================
Development of business competition today makes the role of supply chain management increasingly important to support the company's business, including in the fertilizer industry. PT Petrokimia Gresik as a company engaged in the fertilizer industry, has the largest fertilizer distribution target in the Java-Bali area with product distribution using land transport fleets. To ensure the achievement of the company's targets, it is necessary to have good supply chain performance. This study aims to improve the supply chain performance of the fertilizer industry with a new approach, namely combining supply chain performance measurement models using the SCOR Model with Predictive Analytics using Supervised Machine Learning with random forest (RF) algorithms, support vector machines (SVM) and artificial neural networks ( ANN).
This research has covered descriptive and predictive stages and can be developed in further research by continuing at the prescriptive stage through optimizing improvement proposals, increasing the analysis period for one year with realtime data. From the supply chain performance measurement, it was found that the supply chain responsiveness attribute is a critical performance attribute with the company's KPI parameter that is not achieved is "timeliness of taking orders" with only 37.76% achievement Then analyzed the variables that affect the feature selection, it was found that there are 9 variables that affect the timeliness of retrieval. These nine influential variables were carried out by descriptive analysis and predictive analysis of machine learning classification with the Random Forest (RF) algorithm, Support Vector Machine and Artificial Neural Network. It was found that the RF algorithm has the best prediction rate with an accuracy of 83.91% and an area under curve value. (AUC) of 0.88 (good classification). From the Gini ratio in the RF algorithm, it was found that there are 4 factors that are very significant to the timeliness of collection, namely Qty of delivery orders (POSTO), product type, POSTO order date and shipping costs. Recommendations for improvement are made by making a delay prediction model and the proposed new POSTO order flow, which has an impact on the estimated supply chain performance improvement of 136% and the estimated financial savings of Rp 4.37 billion.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Performa Supply Chain, SCOR Model, Supervised Machine Learning, Supply Chain Performance |
| Subjects: | H Social Sciences > HD Industries. Land use. Labor > HD38.5 Business logistics--Cost effectiveness. Supply chain management. ERP Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) |
| Divisions: | Faculty of Creative Design and Digital Business (CREABIZ) > Technology Management > 61101-(S2) Master Thesis |
| Depositing User: | Ari Primantara |
| Date Deposited: | 02 Aug 2021 07:40 |
| Last Modified: | 13 Sep 2024 06:43 |
| URI: | http://repository.its.ac.id/id/eprint/84692 |
Actions (login required)
 |
View Item |