Forecasting And Visualizing Sales Data For Conveyor And V-Belt Products At Pt Bando Indonesia

Muhammad, Gigih Iman (2025) Forecasting And Visualizing Sales Data For Conveyor And V-Belt Products At Pt Bando Indonesia. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026201082-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy

Abstract

PT Bando Indonesia adalah perusahaan manufaktur yang menjual produk Conveyor dan V-belt. Proyek akhir ini menggunakan model ARIMA (Auto Regressive Integrated Moving Average). Model ARIMA efektif untuk data time series yang stasioner dan data yang telah diubah untuk mencapai stasioneritas. Selain itu, proyek ini mengintegrasikan metode deep learning Long Short-Term Memory (LSTM) untuk mengidentifikasi pola jangka panjang dan ketergantungan waktu pada data time series nonlinier. LSTM memungkinkan model yang lebih fleksibel untuk dinamika pasar dan fluktuasi penjualan yang kompleks. Dengan menganalisis data penjualan historis dan mengidentifikasi pola jangka panjang, Tugas akhir ini bertujuan untuk memprediksi penjualan masa depan dengan lebih akurat dan mengoptimalkan proses produksi. Kombinasi peramalan statistik dan deep learning di PT Bando Indonesia bertujuan untuk meningkatkan akurasi peramalan penjualan dan efektivitas representasi data penjualan. Hasil yang diharapkan meliputi keselarasan strategis penjualan yang lebih baik dan keunggulan kompetitif di sektor otomotif dan industri. Hasil penelitian ini menunjukkan bahwa kombinasi model ARIMA dan LSTM meningkatkan akurasi peramalan penjualan di PT Bando Indonesia. Model ARIMA efektif untuk produk dengan pola penjualan stabil. Sementara itu, model LSTM unggul dalam memodelkan permintaan yang kompleks, nonlinier, dan volatil. Secara keseluruhan, Mean Absolute Percentage Error (MAPE) yang dihasilkan oleh ARIMA berkisar antara 7,87% hingga 23,08%, sedangkan nilai MAPE yang dihasilkan oleh LSTM berkisar antara 8,33% hingga 22,33%. Hasil ini menunjukkan bahwa pemilihan model yang tepat harus didasarkan pada karakteristik pola penjualan masing-masing produk. ARIMA lebih cocok untuk pola yang stabil, sedangkan LSTM lebih efektif untuk pola yang kompleks dan dinamis. Hasil peramalan ini memungkinkan PT Bando Indonesia untuk mengantisipasi perubahan pasar dan menyelaraskan produksi dengan permintaan aktual. Kemampuan peramalan yang ditingkatkan memberikan keunggulan kompetitif bagi perusahaan, memungkinkan respons yang lebih proaktif terhadap kondisi pasar yang dinamis.
======================================================================================================================================================================
PT Bando Indonesia is a manufacturing company that selling Conveyor and V-belt products. This final project employs the ARIMA (Auto Regressive Integrated Moving Average) model. The ARIMA model is effective for stationary time series data and for data that has been transformed to achieve stationarity. Additionally, the project incorporates the deep learning method Long Short-Term Memory (LSTM) to identify long-term patterns and time dependencies in nonlinear time series data. LSTM allows for a more flexible model of market dynamics and complex sales fluctuations. By analyzing historical sales data, considering external influences, and identifying long-term patterns, this thesis aims to predict future sales more accurately and optimize the production process. The combined statistical forecasting and deep learning at PT. Bando Indonesia, aims to improving the accuracy of sales forecasts and the effectiveness of sales data representation. Expected results include better strategic sales alignment, and a competitive advantage in the automotive and industrial sectors. The results of this research indicate that combining ARIMA and LSTM models improves the accuracy of sales forecasts at PT Bando Indonesia. The ARIMA model is effective for products with stable or regular sales patterns. Meanwhile, the LSTM model excels at modelling complex, nonlinear, and volatile demand. Overall, the Mean Absolute Percentage Error (MAPE) by ARIMA ranges from 7.87% to 23.08%, while the MAPE value produced by LSTM ranges from 8.33% to 22.33%. These results suggest that selecting the appropriate model should be based on the characteristics of each product's sales pattern. ARIMA is more suitable for stable patterns, while LSTM is more effective for complex and dynamic patterns. These forecasting results allow PT Bando Indonesia to anticipate market changes, align production with actual demand. The improved forecasting capability gives the company a competitive advantage, enabling a more proactive response to dynamic market conditions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Conveyor, V-Belt, Peramalan Penjualan, Autoregressive Integrated Moving Average (ARIMA), Deep Learning, Long Short-Term Memory (LSTM).==================================================================================================================== Conveyor, V-Belt, Sales Forecasting, Autoregressive Integrated Moving Average (ARIMA), Deep Learning, Long Short-Term Memory (LSTM).
Subjects:	Q Science > QA Mathematics > QA76.9.I52 Information visualization T Technology > T Technology (General) > T174 Technological forecasting
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Gigih Iman Muhammad
Date Deposited:	26 Jul 2025 08:20
Last Modified:	26 Jul 2025 08:20
URI:	http://repository.its.ac.id/id/eprint/122185

Actions (login required)

View Item