| dc.contributor.advisor | Wicaksono, Aditya | |
| dc.contributor.author | FARIDA, RIMA TRIA | |
| dc.date.accessioned | 2025-08-23T03:03:39Z | |
| dc.date.available | 2025-08-23T03:03:39Z | |
| dc.date.issued | 2025 | |
| dc.identifier.uri | http://repository.ipb.ac.id/handle/123456789/170146 | |
| dc.description.abstract | Berdasarkan data produksi terdapat kenaikan reject product pada bulan
Agustus 2024 sebanyak 3.813 reject dengan persentase 2,15% ini melebihi standar
yang telah ditetapkan perusahaan yaitu 0,10%. Akar masalah terjadinya reject
tinggi karena ketidaksesuaian parameter mesin produksi. Tujuan penelitian ini
adalah mengimplementasikan machine learning untuk memprediksi reject product
dengan melakukan perbandingan kinerja algoritma LSTM, RNN, XGBoost, dan
Random Forest, serta menerapkan fitur dashboard. Penelitian ini menggunakan
metodologi CRISP-DM yang digunakan untuk proses analisis data dan proyek data
mining. Perbandingan algoritma dilakukan melalui evaluasi metrik Mean Absolute
Error (MAE) dan Root Mean Squared Error (RMSE). Dari perbandingan algoritma
yang telah dilakukan, hasilnya menunjukkan bahwa LSTM merupakan algoritma
terbaik karena mampu mengenali pola data reject berbasis time series dengan nilai
MAE sebesar 36.92 dan RMSE sebesar 114.47. Dengan demikian, algoritma LSTM
dipilih untuk diimplementasikan dalam sistem prediksi reject product | |
| dc.description.abstract | Based on production data, there was an increase in rejected products in
August 2024, totaling 3,813 rejects with a percentage of 2.15%, which exceeds the
company's established standard of 0.10%. The root cause of the high rejection rate
is the mismatch of production machine parameters. The objective of this study is to
implement machine learning to predict rejected products by comparing the
performance of the LSTM, RNN, XGBoost, and Random Forest algorithms, as well
as applying dashboard features. This study uses the CRISP-DM methodology for
data analysis and data mining projects. The algorithms are compared through the
evaluation of the Mean Absolute Error (MAE) and Root Mean Squared Error
(RMSE) metrics. From the algorithm comparison conducted, the results show that
LSTM is the best algorithm because it can recognize time series-based reject data
patterns with an MAE value of 36.92 and an RMSE value of 114.47. Thus, the
LSTM algorithm was selected for implementation in the reject product prediction
system. | |
| dc.description.sponsorship | | |
| dc.language.iso | id | |
| dc.publisher | IPB University | id |
| dc.title | Perbandingan LSTM, RNN, XGBoost, Random Forest untuk Prediksi Produk Gagal di PT Amerta Indah Otsuka | id |
| dc.title.alternative | Comparison of LSTM, RNN, XGBoost, and Random Forest for Predicting Product Failures at PT Amerta Indah Otsuka | |
| dc.type | Tugas Akhir | |
| dc.subject.keyword | comparison | id |
| dc.subject.keyword | dashboard | id |
| dc.subject.keyword | machine learning | id |
| dc.subject.keyword | time series | id |