Time-series predictive analytics in mining: a systematic review with bibliometric mapping and thematic synthesis (1968-2025)

Purpose This study aims to systematically review the use of time-series predictive analytics in the mining industry, highlighting its role in anticipating critical operational, environmental and geotechnical events through data obtained from sensors, telemetry and remote monitoring technologies. Design/methodology/approach A mixed-methods approach was used, combining bibliometric analysis, text mining and qualitative content synthesis. The review followed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) 2020 protocol to select 135 peer-reviewed articles published between 1968 and 2025, complemented by an ad hoc search to ensure comprehensiveness. The analysis encompassed both traditional statistical forecasting models and contemporary machine learning (ML) and deep learning (DL) techniques. Findings The review reveals a marked increase in research activity since 2022 and identifies two main thematic clusters: (i) operational applications of time series forecasting in underground and open-pit mining, focusing on processes such as haulage, drilling/blasting, ventilation and geotechnical stability; and (ii) ML/DL-based forecasting models, particularly sequential (LSTM/gated recurrent unit (GRU)), attention-based (transformers) and hybrid architectures. Key challenges identified include model transferability, external validation and uncertainty quantification. The study proposes four future research directions to enhance the robustness, interpretability and practical applicability of forecasting models in real-world mining scenarios. Originality/value This study offers the first integrated review of time-series forecasting in mining, combining bibliometric mapping and thematic synthesis from 1968 to 2025. It highlights how traditional and ML/DL models are applied to key mining operations and identifies gaps in model transferability, validation and uncertainty. The findings provide actionable insights for improving predictive analytics in real-world mining contexts.