Ines Wilms
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Ines Wilms has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- Stephan Smeekes
Department of Quantitative Economics Maastricht University
7 shared publications Christophe Croux
KU Leuven - Faculty of Business and Economics (FEB)
5 shared publications- David S. Matteson
Cornell University
4 shared publications - Jakob Raymaekers
Department of Mathematics, University of Antwerp, Antwerp, Belgium
3 shared publications Jeroen Rombouts
Maastricht University Purdue University
3 shared publications- Peter Filzmoser
Institute of Statistics and Mathematical Methods in Economics, TU Wien
3 shared publications
Latest Publications
- Open Access
- 28 October 2026
- Published by EDP Sciences in Journal of Space Weather and Space Climate
- Vol. 16, 38
Performance of radio-based detection for operational monitoring of M5+ class solar flares
Early detection of major solar flares is critical due to their potential to disturb radar and radio systems. Typically, soft X-ray flux is used to monitor and classify solar flares, but since this flux has to be measured in space, it means that its availability itself is dependent on space weather conditions. For this reason, in this paper, we investigated the feasibility of using ground radio observations to monitor major (M5+ class) solar flares. We made use of datasets from the GOES-16 satellite and the Radio Solar Telescope Network in the time range between March 2023 and March 2025. An elastic net-regularized logistic regression model was trained on these data, optimized through a grid search and with incorporated class weighting for class imbalance. It was found that especially higher frequencies (8800 MHz) had a reasonable ability for monitoring and predicting major flares (precision and recall for flare events are 53% and 65%, respectively– implying that roughly one third of flares were not detected – with signals appearing, on average, 3–4 min before the M5 threshold is exceeded). Radio measurements at super high frequencies can thus serve as an alternative method to monitor major solar flaring activity.
- 28 October 2026
- Published by EDP Sciences in Journal of Space Weather and Space Climate
- Vol. 16, 38
Open AccessPerformance of radio-based detection for operational monitoring of M5+ class solar flares
Early detection of major solar flares is critical due to their potential to disturb radar and radio systems. Typically, soft X-ray flux is used to monitor and classify solar flares, but since this flux has to be measured in space, it means that its availability itself is dependent on space weather conditions. For this reason, in this paper, we investigated the feasibility of using ground radio observations to monitor major (M5+ class) solar flares. We made use of datasets from the GOES-16 satellite and the Radio Solar Telescope Network in the time range between March 2023 and March 2025. An elastic net-regularized logistic regression model was trained on these data, optimized through a grid search and with incorporated class weighting for class imbalance. It was found that especially higher frequencies (8800 MHz) had a reasonable ability for monitoring and predicting major flares (precision and recall for flare events are 53% and 65%, respectively– implying that roughly one third of flares were not detected – with signals appearing, on average, 3–4 min before the M5 threshold is exceeded). Radio measurements at super high frequencies can thus serve as an alternative method to monitor major solar flaring activity.
- Open Access
- 8 July 2026
- Published by Elsevier in Econometrics and Statistics
Decomposing co-movements in matrix-valued time series: A pseudo-structural reduced rank approach
- 8 July 2026
- Published by Elsevier in Econometrics and Statistics
Open AccessDecomposing co-movements in matrix-valued time series: A pseudo-structural reduced rank approach
- Open Access
- 20 May 2026
- Published by Taylor & Francis in Journal of Computational and Graphical Statistics
- p. 1-15
Clusterpath Gaussian Graphical Modeling
- 20 May 2026
- Published by Taylor & Francis in Journal of Computational and Graphical Statistics
- p. 1-15
Open AccessClusterpath Gaussian Graphical Modeling
- 21 April 2026
- Published by The R Foundation
FoRecoML: Forecast Reconciliation with Machine Learning
Nonlinear forecast reconciliation with machine learning in cross-sectional (Spiliotis et al. 2021 ), temporal, and cross-temporal (Rombouts et al. 2024 ) frameworks.
- 21 April 2026
- Published by The R Foundation
FoRecoML: Forecast Reconciliation with Machine Learning
Nonlinear forecast reconciliation with machine learning in cross-sectional (Spiliotis et al. 2021 ), temporal, and cross-temporal (Rombouts et al. 2024 ) frameworks.
- Open Access
Autotune: fast, accurate, and automatic tuning parameter selection for Lasso
Least absolute shrinkage and selection operator (Lasso), a popular method for high-dimensional regression, is now used widely for estimating high-dimensional time series models such as the vector autoregression (VAR). Selecting its tuning parameter efficiently and accurately remains a challenge, despite the abundance of available methods for doing so. We propose $\mathsf{autotune}$, a strategy for Lasso to automatically tune itself by optimizing a penalized Gaussian log-likelihood alternately over regression coefficients and noise standard deviation. Using extensive simulation experiments on regression and VAR models, we show that $\mathsf{autotune}$ is faster, and provides better generalization and model selection than established alternatives in low signal-to-noise regimes. In the process, $\mathsf{autotune}$ provides a new estimator of noise standard deviation that can be used for high-dimensional inference, and a new visual diagnostic procedure for checking the sparsity assumption on regression coefficients. Finally, we demonstrate the utility of $\mathsf{autotune}$ on a real-world financial data set. An R package based on C++ is also made publicly available on Github.Open AccessAutotune: fast, accurate, and automatic tuning parameter selection for Lasso
Least absolute shrinkage and selection operator (Lasso), a popular method for high-dimensional regression, is now used widely for estimating high-dimensional time series models such as the vector autoregression (VAR). Selecting its tuning parameter efficiently and accurately remains a challenge, despite the abundance of available methods for doing so. We propose $\mathsf{autotune}$, a strategy for Lasso to automatically tune itself by optimizing a penalized Gaussian log-likelihood alternately over regression coefficients and noise standard deviation. Using extensive simulation experiments on regression and VAR models, we show that $\mathsf{autotune}$ is faster, and provides better generalization and model selection than established alternatives in low signal-to-noise regimes. In the process, $\mathsf{autotune}$ provides a new estimator of noise standard deviation that can be used for high-dimensional inference, and a new visual diagnostic procedure for checking the sparsity assumption on regression coefficients. Finally, we demonstrate the utility of $\mathsf{autotune}$ on a real-world financial data set. An R package based on C++ is also made publicly available on Github. - Open Access
Robust outlier-adjusted mean-shift estimation of state-space models
State-space models (SSMs) provide a flexible framework for modelling time series data, but their reliance on Gaussian error assumptions makes them highly sensitive to outliers. We propose a robust estimation method, ROAMS, that mitigates the influence of additive outliers by introducing shift parameters at each timepoint in the observation equation of the SSM. These parameters allow the model to attribute non-zero shifts to outliers while leaving clean observations unaffected. ROAMS then enables automatic outlier detection, through the addition of a penalty term on the number of flagged outlying timepoints in the objective function, and simultaneous estimation of model parameters. We apply the method to robustly estimate SSMs on both simulated data and real-world animal location-tracking data, demonstrating its ability to produce more reliable parameter estimates than classical methods and other benchmark methods. In addition to improved robustness, ROAMS offers practical diagnostic tools, including BIC curves for selecting tuning parameters and visualising outlier structure. These features make our approach broadly useful for researchers and practitioners working with contaminated time series data.
Open AccessRobust outlier-adjusted mean-shift estimation of state-space models
State-space models (SSMs) provide a flexible framework for modelling time series data, but their reliance on Gaussian error assumptions makes them highly sensitive to outliers. We propose a robust estimation method, ROAMS, that mitigates the influence of additive outliers by introducing shift parameters at each timepoint in the observation equation of the SSM. These parameters allow the model to attribute non-zero shifts to outliers while leaving clean observations unaffected. ROAMS then enables automatic outlier detection, through the addition of a penalty term on the number of flagged outlying timepoints in the objective function, and simultaneous estimation of model parameters. We apply the method to robustly estimate SSMs on both simulated data and real-world animal location-tracking data, demonstrating its ability to produce more reliable parameter estimates than classical methods and other benchmark methods. In addition to improved robustness, ROAMS offers practical diagnostic tools, including BIC curves for selecting tuning parameters and visualising outlier structure. These features make our approach broadly useful for researchers and practitioners working with contaminated time series data.
- Open Access
- 18 November 2025
- Published by Springer Nature in Statistics and Computing
- Vol. 36 (1) , 22
Sparse outlier-robust PCA for multi-source data
Sparse and outlier-robust principal component analysis (PCA) has been a very active field of research recently. Yet, most existing methods apply PCA to a single data set whereas multi-source data—i.e. multiple related data sets requiring joint analysis—arise across many scientific areas. We introduce a novel PCA methodology that simultaneously (i) selects important features, (ii) allows for the detection of global sparse patterns across multiple data sources as well as local source-specific patterns, and (iii) is resistant to outliers. To this end, we develop a regularization problem with a penalty that accommodates global-local structured sparsity patterns, and where an outlier-robust covariance estimator, namely the ssMRCD, is used as plug-in to permit joint, robust analysis across multiple data sources. We provide an efficient implementation of our proposal via the alternating direction method of multipliers and illustrate its practical advantages in simulations and in applications.
- 18 November 2025
- Published by Springer Nature in Statistics and Computing
- Vol. 36 (1) , 22
Open AccessSparse outlier-robust PCA for multi-source data
Sparse and outlier-robust principal component analysis (PCA) has been a very active field of research recently. Yet, most existing methods apply PCA to a single data set whereas multi-source data—i.e. multiple related data sets requiring joint analysis—arise across many scientific areas. We introduce a novel PCA methodology that simultaneously (i) selects important features, (ii) allows for the detection of global sparse patterns across multiple data sources as well as local source-specific patterns, and (iii) is resistant to outliers. To this end, we develop a regularization problem with a penalty that accommodates global-local structured sparsity patterns, and where an outlier-robust covariance estimator, namely the ssMRCD, is used as plug-in to permit joint, robust analysis across multiple data sources. We provide an efficient implementation of our proposal via the alternating direction method of multipliers and illustrate its practical advantages in simulations and in applications.
- 17 October 2025
- Published by The R Foundation
clusterGGM: Sparse Gaussian Graphical Modeling with Variable Clustering
Perform sparse estimation of a Gaussian graphical model (GGM) with node aggregation through variable clustering. Currently, the package implements the clusterpath estimator of the Gaussian graphical model (CGGM) (Touw, Alfons, Groenen & Wilms, 2025; ).
- 17 October 2025
- Published by The R Foundation
clusterGGM: Sparse Gaussian Graphical Modeling with Variable Clustering
Perform sparse estimation of a Gaussian graphical model (GGM) with node aggregation through variable clustering. Currently, the package implements the clusterpath estimator of the Gaussian graphical model (CGGM) (Touw, Alfons, Groenen & Wilms, 2025; ).
- Open Access
MLOps Monitoring at Scale for Digital Platforms
Machine learning models are widely recognized for their strong performance in forecasting. To keep that performance in streaming data settings, they have to be monitored and frequently re-trained. This can be done with machine learning operations (MLOps) techniques under supervision of an MLOps engineer. However, in digital platform settings where the number of data streams is typically large and unstable, standard monitoring becomes either suboptimal or too labor intensive for the MLOps engineer. As a consequence, companies often fall back on very simple worse performing ML models without monitoring. We solve this problem by adopting a design science approach and introducing a new monitoring framework, the Machine Learning Monitoring Agent (MLMA), that is designed to work at scale for any ML model with reasonable labor cost. A key feature of our framework concerns test-based automated re-training based on a data-adaptive reference loss batch. The MLOps engineer is kept in the loop via key metrics and also acts, pro-actively or retrospectively, to maintain performance of the ML model in the production stage. We conduct a large-scale test at a last-mile delivery platform to empirically validate our monitoring framework.
Open AccessMLOps Monitoring at Scale for Digital Platforms
Machine learning models are widely recognized for their strong performance in forecasting. To keep that performance in streaming data settings, they have to be monitored and frequently re-trained. This can be done with machine learning operations (MLOps) techniques under supervision of an MLOps engineer. However, in digital platform settings where the number of data streams is typically large and unstable, standard monitoring becomes either suboptimal or too labor intensive for the MLOps engineer. As a consequence, companies often fall back on very simple worse performing ML models without monitoring. We solve this problem by adopting a design science approach and introducing a new monitoring framework, the Machine Learning Monitoring Agent (MLMA), that is designed to work at scale for any ML model with reasonable labor cost. A key feature of our framework concerns test-based automated re-training based on a data-adaptive reference loss batch. The MLOps engineer is kept in the loop via key metrics and also acts, pro-actively or retrospectively, to maintain performance of the ML model in the production stage. We conduct a large-scale test at a last-mile delivery platform to empirically validate our monitoring framework.
- 1 March 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 36 (1) , 552-571
Fast Forecasting of Unstable Data Streams for On-Demand Service Platforms
Practice- and policy-oriented The success of on-demand service platforms crucially hinges upon their ability to make fast and accurate demand forecasts so that its workers are always at the right time and location to serve customers promptly. Yet demand forecasting is challenging for several reasons. First, demand data are typically released as high-frequency streaming time series, which requires an algorithm that has a fast processing time. Second, a digital platform often operates in many different geographic regions, thereby giving rise to a large heterogeneous geographical collection of high-frequency demand streams that need to be forecast and requiring a scalable algorithm. Third, a platform business usually operates in an unstable, rapidly changing environment and faces irregular growth patterns, which requires agility when forecasting demand because slow reactions to such instabilities causes forecast performance to break down. We offer a novel forecast framework called fast forecasting of unstable data streams that is fast and scalable and automatically assesses changing environments without human intervention. We test our framework on a unique data set from a leading European on-demand delivery platform and a U.S. bicycle sharing system and find strong (i) forecast performance gains, (ii) financial gains, and (ii) computing time reduction from using our framework against several industry benchmarks.
- 1 March 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 36 (1) , 552-571
Fast Forecasting of Unstable Data Streams for On-Demand Service Platforms
Practice- and policy-oriented The success of on-demand service platforms crucially hinges upon their ability to make fast and accurate demand forecasts so that its workers are always at the right time and location to serve customers promptly. Yet demand forecasting is challenging for several reasons. First, demand data are typically released as high-frequency streaming time series, which requires an algorithm that has a fast processing time. Second, a digital platform often operates in many different geographic regions, thereby giving rise to a large heterogeneous geographical collection of high-frequency demand streams that need to be forecast and requiring a scalable algorithm. Third, a platform business usually operates in an unstable, rapidly changing environment and faces irregular growth patterns, which requires agility when forecasting demand because slow reactions to such instabilities causes forecast performance to break down. We offer a novel forecast framework called fast forecasting of unstable data streams that is fast and scalable and automatically assesses changing environments without human intervention. We test our framework on a unique data set from a leading European on-demand delivery platform and a U.S. bicycle sharing system and find strong (i) forecast performance gains, (ii) financial gains, and (ii) computing time reduction from using our framework against several industry benchmarks.