PENGGUNAAN TEKNIK DATA MINING UNTUK PERAMALAN PERMINTAAN SEMEN DI PT X
PT X is a national company (BUMN) that produces construction materials in the form of various types of cement and its derivatives needed for construction. This company is a holding company that has many production plants located in various regions in Indonesia and abroad. One of the production pl...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/54201 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | PT X is a national company (BUMN) that produces construction materials in the
form of various types of cement and its derivatives needed for construction. This
company is a holding company that has many production plants located in various
regions in Indonesia and abroad. One of the production plants of PT X is located
in Tuban Regency, East Java Province. The PT X factory, which is located in Tuban
Regency, East Java Province, has a production target of four to five million tons of
cement per year. To achieve this production target, PT X requires a better
production planning process. To support its production planning, PT X requires a
production planning system that can assist in aggregate planning and disaggregate
planning for cement production.
Departemen Perencanaan Produksi dan Evaluasi PT X requires a cement demand
forecasting model. Therefore, this study was conducted to develop a demand
forecasting model for cement. There are six forecasting models developed in this
research. Three forecasting models consisting of ARIMA, XGBoost, and LSTM are
used for time series forecasting, while the other three models consisting of the
Croston method, the TSB method, and the modified SBA method are used for
forecasting intermittent demand. The deep learning model with LSTM architecture
was chosen because it achieved the best score on five time series forecasting models
with root mean square error (RMSE) score of 8082.30, 3679.61, 5753.43, 3657.19,
dan 2248.68. And mean absolute percentage error (MAPE) score of 18.63%,
29.42%, 19.26%, 30.32%, dan 9.67%. For intermittent demand, the TSB model was
chosen for three intermittent demand forecasting models with the best RMSE score
of 107,82, 3555.34, and 3334.93. The Croston model was chosen for intermitent
data model with the best RMSE score of 3914.23.
|
|---|