Machine Learning Methods for Forecasting Intermittent Tin Ore Production - Vol. 8 No. 5

Aug 5, 2026

Upload by Liam

Effective production forecasting is critical for resource planning and management in the mining industry, especially when tin ore output is intermittent. Cutter Section Dredges (CSD) tin ore production can fluctuate and include frequent zero-production days, making classical forecasting less suitable. This study proposes a serial combination of machine learning classification and forecasting to predict the next day’s production, using Random Forest and CatBoost for operational status detection, then CatBoost and Bi-LSTM for magnitude forecasting during operational periods. Results show improved accuracy for intermittent time series.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Abstract

1. Introduction

Tin ore production and forecasting needs

Intermittency and zero values challenge

Prior forecasting methods

Deep learning and interpretable machine learning

2. Rese

Accredited SINTA 2 Ranking Decree of the Director General of Higher Education, Research, and Technology, No. 158/E/KPT/2021 Validity period from Volume 5 Number 2 of 2021 to Volume 10 Number 1 of 2026 Published online at: [http://jurnal.iaii.or.id](http://jurnal.iaii.or.id) JURNAL RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol. 8 No. 5 (2024) 644-650 e-ISSN: 2580-0760 Machine Learning Methods for Forecasting Intermittent Tin Ore Production Nabila Dhia Alifa Rahmah 1*, Budhi Handoko2, Anindya Apriliyanti Pravitasari3 1,2,3Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Padjadjaran, Bandung, Indonesia [1](1 nabila17035@mail.unpad.ac.id)[ nabila17035@mail.unpad.ac.id](1 nabila17035@mail.unpad.ac.id), [2](2budhi.handoko@unpad.ac.id)[budhi.handoko@unpad.ac.id](2budhi.handoko@unpad.ac.id), [3](3 anindya.apriliyanti@unpad.ac.id)[ anindya.apriliyanti@unpad.ac.id](3 anindya.apriliyanti@unpad.ac.id) Abstract Effective production forecasting is important for resource planning and management in the mining industry. Tin oreproduction from Cutter Section Dredges (CSD) may fluctuate due to a variety offactors, in which there are periods when the production is zero. This study compares various combinations of machine learning-based classification and forecasting to predict future tin oreproduction values, which have not been found in previous studies. The presence of zero values in the forecast in the next day's tin ore production forecast is addressed by combining classification and forecasting techniques. Random Forest and CatBoost classification techniques are used to determine the next day's CSD production operating status. Then, for each time point when the CSD is operational, a forecasting model is created using CatBoost and Bi-LSTM. This study's findings show that a serial combination of the Random Forest classification method and CatBoost forecasting can produce accurate tin oreproduction forecasts for the selected CSD (RMSE = 0.271, MAE = 0.179, MAE = 0.730, F1-score = 0,80). This study demonstrates how a serial combination of classification and forecasting models can improve the accuracy and efficiency of production forecasting for intermittent time series data. Keywords: forecasting; classification, machine learning; mining; CatBoost How to Cite: N. D. A. Rahmah, B. Handoko, and A. A. Pravitasari,“Machine Learning Methods for Forecasting Intermittent Tin Ore Production”, J. RESTI (Rekayasa Sist. Teknol. Inf.) , vol. 8, no. 5, pp. 644-650, Oct. 2024. DOI: [https://doi.org/10.29207/resti.v8i5.5990](https://doi.org/10.29207/resti.v8i5.5990) 1. Introduction Tin ore production plays a crucial part in the mining industry as it provides raw materials needed for a variety of industries. Effective forecasting of tin oreproduction aims to provide an overview of future trends in tin ore production and may aid in resource planning and management. Offshore tin ore production using Cutter Section Dredges (CSDs) varies throughout the year due to changing weather, equipment and operational factors, and other field conditions, as well as unpredictable docking times that can result in zero production on some days. Up to 30.3% of the production data used in this study contain zero values. Previous studies have been done on the prediction and forecasting of various mining outcomes, employing a variety of machine-learning techniques that combine historical production data with specific operational data for each mine type [1] - [8] . However, discussions on production forecasting, particularly for tin ore, are difficult to find. Several studies have also examined forecasting data with a large number of zero values, which is common in the context of intermittent demand forecasts. [9], [10] take an interesting approach to overcome this problem by splitting it into two tasks: classifying the presence/absence of demand and predicting the magnitude of demand, which can provide

FAQs

Why is forecasting tin ore production difficult for Cutter Section Dredges (CSD)?

CSD tin ore production varies due to weather, equipment and operational factors, and docking times that can cause zero output on some days.

How does the study address zero values in next-day tin ore production forecasting?

It separates the problem into two serial phases: classification to predict whether CSD is operational, then forecasting only for time points when production is non-zero.

What models are used in the combined approach?

Random Forest and CatBoost are used for the classification of next-day operational status, while CatBoost and Bi-LSTM are used for forecasting production at operational time steps.

Aug 5, 2026

Upload by Liam