Effective public engagement with climate change is central to advancing sustainability goals, yet the factors shaping audience responses to climate-related digital content remain insufficiently understood. This study investigates how presenter identity, message framing, and interaction structure influence audience engagement with climate change videos on YouTube. Using a mixed-methods approach, we analysed 129 English-language YouTube videos and their associated user comments, combining manual coding of video attributes with natural language processing and supervised machine learning to analyse comment sentiment. A binary logistic regression model was used to predict positive versus negative audience attitudes at the video level, with chi-square tests employed as supporting analyses. Results indicate that videos presented by scientists are significantly more likely to elicit positive audience attitudes than those presented by politicians or other public figures. Solution-focused framing is strongly associated with positive engagement, while blame-oriented framing is associated with negative responses. Additionally, threaded comment discussions show a higher proportion of positive attitudes than independent comments, suggesting that conversational interaction enhances constructive engagement. These findings highlight the importance of expertise-based communication, solution-oriented narratives, and interactive discourse in digital sustainability communication. The study contributes both methodological tools and practical insights for designing climate change communication strategies that foster informed and constructive public engagement. © 2026 Aharonson, Christodoulou, Karpasitis, Joselowitz, Nowaczyk, Lazebnik and Iordanou.

Efficient management of spare parts inventory is crucial in the automotive aftermarket, where demand is highly intermittent, and uncertainty drives substantial cost and service risks. Forecasting is therefore central, but the quality of a forecasting model should be judged not by statistical accuracy (e.g., MAE, RMSE, IAE) but rather by its impact on key operational performance indicators (KPIs), such as total cost and service level. Yet most existing work evaluates models exclusively using accuracy metrics, and the relationship between these metrics and operational KPIs remains poorly understood. To address this gap, we propose a decision-centric simulation software framework that enables the systematic evaluation of forecasting models in realistic inventory management settings. The framework comprises: (i) a synthetic demand generator tailored to spare-parts demand characteristics, (ii) a flexible forecasting module that can host arbitrary predictive models, and (iii) an inventory control simulator that consumes the forecasts and computes, based on selected inventory control policy, operational KPIs. This closed-loop setup enables practitioners and researchers to evaluate models not only in terms of statistical error but also in terms of their downstream implications for inventory decisions. Using a wide range of simulation scenarios, we show that improvements in conventional accuracy metrics do not necessarily translate into better operational performance, and that models with similar statistical error profiles can induce markedly different cost–service trade-offs. We analyze these discrepancies to characterize how specific aspects of forecast performance affect inventory outcomes and to derive actionable guidance for model selection. Overall, the framework operationalizes the link between demand forecasting and inventory management, shifting evaluation from purely predictive accuracy towards operational relevance in the automotive aftermarket and related domains. An open-source implementation of the software, including all experimental results, is available at https://github.com/caisr-hh/TruckParts-Demand-Inventory-Simulator/releases/tag/IDA_2026. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Deep learning has demonstrated success in domains such as vision and speech, largely due to its ability to learn hierarchical feature representations via backpropagation. However, tree-based models, such as Random Forest and XGBoost, remain dominant for tabular data, often, but not always, outperforming deep Artificial Neural Networks (ANNs) while requiring less computational resources. It has been demonstrated that neural networks are particularly effective for data that is less heavily preprocessed and that contains important yet complex feature dependencies. This paper introduces Deep Decision Forest (DDF), a multilayer decision tree ensemble that bridges this gap by incorporating a feedback mechanism analogous to backpropagation. We postulate that such a mechanism will enable tree-based models to capture hierarchical feature representations to a degree that was previously impossible. Each layer of trees produces an output vector that serves as the input features for subsequent layers, and final predictions are obtained through majority voting in the last layer. One key difference between tree-based models and ANNs is that, instead of making minor adjustments to all the numeric weights at once, DDF first identifies the most underperforming features across layers, and selectively retrains only the corresponding trees using specialised improvement datasets. Experiments on seven benchmark datasets demonstrate that DDF consistently outperforms a standard Decision Tree, performs as well as or better than Random Forest, and achieves competitive accuracy compared to Deep Forest. These experiments demonstrate that integrating elements inspired by deep learning into tree ensembles is both feasible and effective, offering a new hybrid approach for tabular learning. All the code and experiments are available open source at: https://github.com/caisr-hh/Deep-Decision-Forest. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Accurate energy consumption prediction is crucial for optimizing the operation of electric commercial heavy-duty vehicles, e.g., route planning for charging. Moreover, understanding why certain predictions are cast is paramount for such a predictive model to gain user trust and be deployed in practice. Since commercial vehicles operate differently as transportation tasks, ambient, and drivers vary, a heterogeneous population is expected when building an AI system for forecasting energy consumption. The dependencies between the input features and the target values are expected to also differ across sub-populations. One well-known example of such a statistical phenomenon is Simpson’s paradox. In this paper, we illustrate that such a setting poses a challenge for existing XAI methods that produce global feature statistics, e.g., LIME or SHAP, causing them to yield misleading results. We demonstrate a potential solution by training multiple regression models on subsets of data via a divide-and-conquer approach. It not only leads to superior regression performance but also more relevant and consistent LIME explanations. Given that the employed groupings correspond to relevant sub-populations, the associations between the input features and the target values are consistent within each cluster but different across clusters. Experiments on both synthetic and real-world datasets show that such splitting of a complex problem into simpler ones yields better regression performance and interpretability. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

We propose a new sequential decision-making setting, combining key aspects of two established online learning problems with bandit feedback. The optimal action to play at any given moment is contingent on an underlying changing state that is not directly observable by the agent. Each state is associated with a context distribution, possibly corrupted, allowing the agent to identify the state. Furthermore, states evolve in a Markovian fashion, providing useful information to estimate the current state via state history. In the proposed problem setting, we tackle the challenge of deciding on which of the two sources of information the agent should base its action selection. We present an algorithm that uses a referee to dynamically combine the policies of a contextual bandit and a multi-armed bandit. We capture the time-correlation of states through iteratively learning the action-reward transition model, allowing for efficient exploration of actions. Our setting is motivated by adaptive mobile health (mHealth) interventions. Users transition through different, time-correlated, but only partially observable internal states, determining their current needs. The side information associated with each internal state might not always be reliable, and standard approaches solely rely on the context risk of incurring high regret. Similarly, some users might exhibit weaker correlations between subsequent states, leading to approaches that solely rely on state transitions risking the same. We analyze our setting and algorithm in terms of regret lower bound and upper bounds and evaluate our method on simulated medication adherence intervention data and several real-world data sets, showing improved empirical performance compared to several popular algorithms. © The Author(s) 2024.

Statistical modeling of multivariate time-series data poses significant challenges due to their high dimensionality and complex inter-variable relationships. Reliable forecasts or anomaly detection on these datasets require capturing such relationships within and between the features. While traditional deep learning architectures are good at capturing temporal non-linear patterns within features, they are less efficient at modeling inter-variable relationships explicitly structured as graphs-a capability where Graph Neural Networks (GNNs) excel. Inspired by the success of GNNs, Graph Deviation Network (GDN) was originally proposed for anomaly detection on industrial multivariate time-series data. After proving its merits through experiments with real-world data, GDN gained significant popularity in the research community, claiming to learn the hidden graph structure in any multivariate time-series data. Various modifications to GDN were proposed over the years, but essentially all of them kept its Graph Structure Learning (GSL) module intact. However, until now, this module has never been rigorously evaluated. This work scrutinizes the contribution of the GSL module. Our experiments reveal that the graph learned by GSL is relatively ineffective, and the key to the overall performance achieved by GDN lies almost entirely in the downstream Graph Attention Network (GAT) module. We hope our findings will garner attention for further development of the GSL module of GDN, whose fidelity can improve the performance of GDN variants. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Balancing performance and scalability is a major concern when developing robust ML models for diverse, big-data scenarios, such as predicting demand for a number of products across multiple locations. The two mutually opposite approaches are to use a single ML model for maximizing scalability, often at the expense of performance, or to use a specialized model for each specific use case, which is often prohibitive in terms of computational costs. In this paper, we propose to balance those two approaches using our methods of model clustering and grouping. We achieve the performance level of a single use-case model while preserving the global scalability of the solution. In our experiments, we use a publicly available demand forecasting dataset as a use case. We develop and train baseline shallow ML models and DL models for both maximizing performance and scalability. Then, we showcase a desirable balance that can be achieved using our proposed methods, one that outperforms both shallow ML and specific use-case models. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

This paper introduces a novel transfer learning adapter, the Bridged Attention Module (BAM), designed to enhance the performance of Spatial-Temporal Graph Convolutional Networks (ST-GCN) in data-limited forecasting scenarios. BAM improves fine-tuning efficiency by jointly capturing spatial and temporal dependencies, optimizing information flow, and significantly reducing the number of trainable parameters while preserving model accuracy. Experimental evaluations demonstrate that the BAM-enhanced ST-GCN consistently achieves competitive accuracy and, in some cases, surpasses traditional fine-tuning methods, even with limited data. The effectiveness of this approach is validated using electric vehicle (EV) charging station occupancy forecasting, highlighting the practical utility of BAM. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Heavy-duty battery electric vehicles rely on large and complex energy storage systems (ESS), composed of multiple battery modules, whose individual health and reliability are critical to vehicle performance and safety. This study applies an unsupervised anomaly detection framework, COSMO (Consensus Self-Organizing Models), to a naturalistic real-world dataset collected during routine operations of in-service heavy-duty vehicles. We extend the baseline COSMO by incorporating causal discovery algorithms to help detect early signs of faults in ESS across heterogeneous missions and external conditions. On-board sensors data is collected as a multivariate time series, including information such as voltage, current, temperature, state of charge, etc. Given the wide range of applications of heavy-duty vehicles, these signals typically exhibit extreme variability even under normal operation, making anomaly detection challenging. Causal graph discovery allows us to acquire latent structures that capture the underlying relationships among these influential features. The resulting learned causal graphs, for each battery module, serve as a more consistent representation that captures each battery module’s usage and behavior over time. Since battery modules within the same ESS are expected to behave similarly under comparable operating conditions, COSMO models them as a homogeneous group. We then mark as anomalous modules that are identified to exhibit causal graph representations deviating markedly from the consensus.

Contextual anomaly detection aims to identify objects that are anomalous only within specific contexts, while appearing normal otherwise. However, most existing methods are limited to a single context defined by user-specified features. In practice, identifying the right context is not trivial, even for domain experts. Moreover, for high-dimensional data, the notion of meaningful contexts that can unveil anomalies becomes substantially more complex. For instance, multiple useful contexts can often capture different phenomena. In this work, we introduce ConQuest, a new unsupervised contextual anomaly detection approach that automatically discovers and incorporates multiple contexts useful for detecting and interpreting anomalies. Through experiments on 25 datasets, we show that ConQuest outperforms various state-of-the-art methods. We also demonstrate its benefits in terms of increased direct interpretability. © The Author(s) 2024.