arXiv subjects

Adam M. Sykulski

Publications and source records attributed to Adam M. Sykulski.

At least 19 recordsLinked to original sources

arXiv2026

A Critical Audit of Spatiotemporal Forecasting Benchmark Datasets and Baselines

Graph neural networks (GNNs) are routinely employed for short-range forecasting on multivariate time series with a spatial graph structure. Despite the availability of many alternative datasets, method innovations within this domain are predominantly assessed against a rather limited set of benchmark datasets, most notably Chickenpox, PedalMe, WikiMaths, METR-LA, and PEMS-BAY. The evaluation protocols contain baselines spanning from historical averages to classical machine learning approaches. These baselines often show competitive performance compared to GNNs. In the present work, we take a step back and analyse the benchmark datasets via classical time series methods to uncover why spatially-unaware linear models pose a stronger competitor than previously reported, casting further doubt on the discriminative reliability of the aforementioned widely adopted datasets. Our statistical analysis provides a toolset for identifying significant spatial and temporal correlations, while revealing a structural bias introduced by first-order differenced datasets. We therefore recommend reducing the over-reliance on such datasets for method comparison, and instead advocate for more rigorous statistical evaluation. By applying the results of our analysis to a simple hybrid model, we show how our methodology can lead to novel ways of developing GNN models

cs.LG↗

arXiv2025

The Whittle likelihood for mixed models with application to groundwater level time series

Understanding the processes that influence groundwater levels is crucial for forecasting and responding to hazards such as groundwater droughts. Mixed models, which combine a fixed mean, expressed using independent predictors, with autocorrelated random errors, are used for inference, forecasting and filling in missing values in groundwater level time series. Estimating parameters of mixed models using maximum likelihood has high computational complexity. For large datasets, this leads to restrictive simplifying assumptions such as fixing certain free parameters in practical implementations. In this paper, we propose a method to jointly estimate all parameters of mixed models using the Whittle likelihood, a frequency-domain quasi-likelihood. Our method is robust to missing and non-Gaussian data and can handle much larger data sizes. We demonstrate the utility of our method both in a simulation study and with real-world data, comparing against maximum likelihood and an alternative two-stage approach that estimates fixed and random effect parameters separately.

stat.ME↗

arXiv2025

Causal tail coefficient for compound extremes in multivariate time series

Extreme events are often multivariate in nature. A compound extreme occurs when a combination of variables jointly produces a significant impact, even if individual components are not necessarily marginally extreme. Compound extremes have been observed across a wide range of domains, including space weather, climate, and environmental science. For example, heavy rainfall sustained over consecutive days can impose cumulative stress on urban drainage systems, potentially resulting in flooding. However, most existing methods for detecting extremal causality focus primarily on individual extreme values and lack the flexibility to capture causal relationships between compound extremes. This work introduces a novel framework for detecting causal dependencies between extreme events, including compound extremes. We introduce the compound causal tail coefficient that captures the extremal dependance of compound events between pairs of stationary time series. Based on a consistent estimator of this coefficient, we develop a bootstrap hypothesis test to evaluate the presence and direction of causal relationships. Our method can accommodate nonlinearity and latent confounding variables. We demonstrate the effectiveness of our method by establishing theoretic properties and through simulation studies and an application to space-weather data.

stat.ME↗

arXiv2025

SplitWise Regression: Stepwise Modeling with Adaptive Dummy Encoding

Capturing nonlinear relationships without sacrificing interpretability remains a persistent challenge in regression modeling. We introduce SplitWise, a novel framework that enhances stepwise regression. It adaptively transforms numeric predictors into threshold-based binary features using shallow decision trees, but only when such transformations improve model fit, as assessed by the Akaike Information Criterion (AIC) or Bayesian Information Criterion (BIC). This approach preserves the transparency of linear models while flexibly capturing nonlinear effects. Implemented as a user-friendly R package, SplitWise is evaluated on both synthetic and real-world datasets. The results show that it consistently produces more parsimonious and generalizable models than traditional stepwise and penalized regression techniques.

cs.LG↗

arXiv2024

Navigating Challenges in Spatio-temporal Modelling of Antarctic Krill Abundance: Addressing Zero-inflated Data and Misaligned Covariates

Antarctic krill (Euphausia superba) are among the most abundant species on our planet and serve as a vital food source for many marine predators in the Southern Ocean. In this paper, we utilise statistical spatio-temporal methods to combine data from various sources and resolutions, aiming to model krill abundance. Our focus lies in fitting the model to a dataset comprising acoustic measurements of krill biomass. To achieve this, we integrate climate covariates obtained from satellite imagery and from drifting surface buoys (also known as drifters). Additionally, we use sparsely collected krill biomass data obtained from net fishing efforts (KRILLBASE) for validation. However, integrating these multiple heterogeneous data sources presents significant modelling challenges, including spatio-temporal misalignment and inflated zeros in the observed data. To address these challenges, we fit a Hurdle-Gamma model to jointly describe the occurrence of zeros and the krill biomass for the non-zero observations, while also accounting for misaligned and heterogeneous data sources, including drifters. Therefore, our work presents a comprehensive framework for analysing and predicting krill abundance in the Southern Ocean, leveraging information from various sources and formats. This is crucial due to the impact of krill fishing, as understanding their distribution is essential for informed management decisions and fishing regulations aimed at protecting the species.

stat.AP↗

arXiv2024

Isotropy testing in spatial point patterns: nonparametric versus parametric replication under misspecification

Several hypothesis testing methods have been proposed to validate the assumption of isotropy in spatial point patterns. A majority of these methods are characterised by an unknown distribution of the test statistic under the null hypothesis of isotropy. Parametric approaches to approximating the distribution involve simulation of patterns from a user-specified isotropic model. Alternatively, nonparametric replicates of the test statistic under isotropy can be used to waive the need for specifying a model. In this paper, we first present a general framework which allows for the integration of a selected nonparametric replication method into isotropy testing. We then conduct a large simulation study comprising application-like scenarios to assess the performance of tests with different parametric and nonparametric replication methods. In particular, we explore distortions in test size and power caused by model misspecification, and demonstrate the advantages of nonparametric replication in such scenarios.

stat.ME↗

arXiv2023

Debiasing Welch's Method for Spectral Density Estimation

Welch's method provides an estimator of the power spectral density that is statistically consistent. This is achieved by averaging over periodograms calculated from overlapping segments of a time series. For a finite length time series, while the variance of the estimator decreases as the number of segments increase, the magnitude of the estimator's bias increases: a bias-variance trade-off ensues when setting the segment number. We address this issue by providing a novel method for debiasing Welch's method which maintains the computational complexity and asymptotic consistency, and leads to improved finite-sample performance. Theoretical results are given for fourth-order stationary processes with finite fourth-order moments and absolutely convergent fourth-order cumulant function. The significant bias reduction is demonstrated with numerical simulation and an application to real-world data. Our estimator also permits irregular spacing over frequency and we demonstrate how this may be employed for signal compression and further variance reduction. Code accompanying this work is available in R and python.

stat.ME↗

arXiv2022

Analysing and visualising bike-sharing demand with outliers

Bike-sharing is a popular component of sustainable urban mobility. It requires anticipatory planning, e.g. of station locations and inventory, to balance expected demand and capacity. However, external factors such as extreme weather or glitches in public transport, can cause demand to deviate from baseline levels. Identifying such outliers keeps historic data reliable and improves forecasts. In this paper we show how outliers can be identified by clustering stations and applying a functional depth analysis. We apply our analysis techniques to the Washington D.C. Capital Bikeshare data set as the running example throughout the paper, but our methodology is general by design. Furthermore, we offer an array of meaningful visualisations to communicate findings and highlight patterns in demand. Last but not least, we formulate managerial recommendations on how to use both the demand forecast and the identified outliers in the bike-sharing planning process.

stat.AP↗

arXiv2022

A multivariate pseudo-likelihood approach to estimating directional ocean wave models

Ocean buoy data in the form of high frequency multivariate time series are routinely recorded at many locations in the world's oceans. Such data can be used to characterise the ocean wavefield, which is important for numerous socio-economic and scientific reasons. This characterisation is typically achieved by modelling the frequency-direction spectrum, which decomposes spatiotemporal variability by both frequency and direction. State-of-the-art methods for estimating the parameters of such models do not make use of the full spatiotemporal content of the buoy observations due to unnecessary assumptions and smoothing steps. We explain how the multivariate debiased Whittle likelihood can be used to jointly estimate all parameters of such frequency-direction spectra directly from the recorded time series. When applied to North Sea buoy data, debiased Whittle likelihood inference reveals smooth evolution of spectral parameters over time. We discuss challenging practical issues including model misspecification, and provide guidelines for future application of the method.

stat.AP↗

arXiv2021

Multivariate Probabilistic Regression with Natural Gradient Boosting

Many single-target regression problems require estimates of uncertainty along with the point predictions. Probabilistic regression algorithms are well-suited for these tasks. However, the options are much more limited when the prediction target is multivariate and a joint measure of uncertainty is required. For example, in predicting a 2D velocity vector a joint uncertainty would quantify the probability of any vector in the plane, which would be more expressive than two separate uncertainties on the x- and y- components. To enable joint probabilistic regression, we propose a Natural Gradient Boosting (NGBoost) approach based on nonparametrically modeling the conditional parameters of the multivariate predictive distribution. Our method is robust, works out-of-the-box without extensive tuning, is modular with respect to the assumed target distribution, and performs competitively in comparison to existing approaches. We demonstrate these claims in simulation and with a case study predicting two-dimensional oceanographic velocity data. An implementation of our method is available at https://github.com/stanfordmlgroup/ngboost.

stat.ML↗

arXiv2021

Outlier detection in network revenue management

This paper presents an automated approach for providing ranked lists of outliers in observed demand to support analysts in network revenue management. Such network revenue management, e.g. for railway itineraries, needs accurate demand forecasts. However, demand outliers across or in parts of a network complicate accurate demand forecasting, and the network structure makes such demand outliers hard to detect. We propose a two-step approach combining clustering with functional outlier detection to identify outlying demand from network bookings observed on the leg level. The first step clusters legs to appropriately partition and pool booking patterns. The second step identifies outliers within each cluster and uses a novel aggregation method across legs to create a ranked alert list of affected instances. Our method outperforms analyses that consider leg data without regard for network implications and offers a computationally efficient alternative to storing and analysing all data on the itinerary level, especially in highly-connected networks where most customers book multi-leg products. A simulation study demonstrates the robustness of the approach and quantifies the potential revenue benefits from adjusting demand forecasts for offer optimisation. Finally, we illustrate the applicability based on empirical data obtained from Deutsche Bahn.

physics.soc-ph↗

arXiv2020

Separating Mesoscale and Submesoscale Flows from Clustered Drifter Trajectories

Drifters deployed in close proximity collectively provide a unique observational data set with which to separate mesoscale and submesoscale flows. In this paper we provide a principled approach for doing so by fitting observed velocities to a local Taylor expansion of the velocity flow field. We demonstrate how to estimate mesoscale and submesoscale quantities that evolve slowly over time, as well as their associated statistical uncertainty. We show that in practice the mesoscale component of our model can explain much first and second-moment variability in drifter velocities, especially at low frequencies. This results in much lower and more meaningful measures of submesoscale diffusivity, which would otherwise be contaminated by unresolved mesoscale flow. We quantify these effects theoretically via computing Lagrangian frequency spectra, and demonstrate the usefulness of our methodology through simulations as well as with real observations from the LatMix deployment of drifters. The outcome of this method is a full Lagrangian decomposition of each drifter trajectory into three components that represent the background, mesoscale, and submesoscale flow.

physics.ao-ph↗

arXiv2020

Estimating the parameters of ocean wave spectra

Wind-generated waves are often treated as stochastic processes. There is particular interest in their spectral density functions, which are often expressed in some parametric form. Such spectral density functions are used as inputs when modelling structural response or other engineering concerns. Therefore, accurate and precise recovery of the parameters of such a form, from observed wave records, is important. Current techniques are known to struggle with recovering certain parameters, especially the peak enhancement factor and spectral tail decay. We introduce an approach from the statistical literature, known as the de-biased Whittle likelihood, and address some practical concerns regarding its implementation in the context of wind-generated waves. We demonstrate, through numerical simulation, that the de-biased Whittle likelihood outperforms current techniques, such as least squares fitting, both in terms of accuracy and precision of the recovered parameters. We also provide a method for estimating the uncertainty of parameter estimates. We perform an example analysis on a data-set recorded off the coast of New Zealand, to illustrate some of the extra practical concerns that arise when estimating the parameters of spectra from observed data.

stat.AP↗

arXiv2020

Nonparametric Time Series Summary Statistics for High-Frequency Accelerometry Data from Individuals with Advanced Dementia

Accelerometry data has been widely used to measure activity and the circadian rhythm of individuals across the health sciences, in particular with people with advanced dementia. Modern accelerometers can record continuous observations on a single individual for several days at a sampling frequency of the order of one hertz. Such rich and lengthy data sets provide new opportunities for statistical insight, but also pose challenges in selecting from a wide range of possible summary statistics, and how the calculation of such statistics should be optimally tuned and implemented. In this paper, we build on existing approaches, as well as propose new summary statistics, and detail how these should be implemented with high frequency accelerometry data. We test and validate our methods on an observed data set from 26 recordings from individuals with advanced dementia and 14 recordings from individuals without dementia. We study four metrics: Interdaily stability (IS), intradaily variability (IV), the scaling exponent from detrended fluctuation analysis (DFA), and a novel nonparametric estimator which we call the proportion of variance (PoV), which calculates the strength of the circadian rhythm using spectral density estimation. We perform a detailed analysis indicating how the time series should be optimally subsampled to calculate IV, and recommend a subsampling rate of approximately 5 minutes for the dataset that has been studied. In addition, we propose the use of the DFA scaling exponent separately for daytime and nighttime, to further separate effects between individuals. We compare the relationships between all these methods and show that they effectively capture different features of the time series.

stat.AP↗

arXiv2020

Estimating the travel time and the most likely path from Lagrangian drifters

We provide a novel methodology for computing the most likely path taken by drifters between arbitrary fixed locations in the ocean. We also provide an estimate of the travel time associated with this path. Lagrangian pathways and travel times are of practical value not just in understanding surface velocities, but also in modelling the transport of ocean-borne species such as planktonic organisms, and floating debris such as plastics. In particular, the estimated travel time can be used to compute an estimated Lagrangian distance, which is often more informative than Euclidean distance in understanding connectivity between locations. Our methodology is purely data-driven, and requires no simulations of drifter trajectories, in contrast to existing approaches. Our method scales globally and can simultaneously handle multiple locations in the ocean. Furthermore, we provide estimates of the error and uncertainty associated with both the most likely path and the associated travel time.

stat.AP↗

arXiv2020

The Elliptical Ornstein-Uhlenbeck Process

We introduce the elliptical Ornstein-Uhlenbeck (OU) process, which is a generalisation of the well-known univariate OU process to bivariate time series. This process maps out elliptical stochastic oscillations over time in the complex plane, which are observed in many applications of coupled bivariate time series. The appeal of the model is that elliptical oscillations are generated using one simple first order stochastic differential equation (SDE), whereas alternative models require more complicated vectorised or higher order SDE representations. The second useful feature is that parameter estimation can be performed semi-parametrically in the frequency domain using the Whittle Likelihood. We determine properties of the model including the conditions for stationarity, and the geometrical structure of the elliptical oscillations. We demonstrate the utility of the model by measuring periodic and elliptical properties of Earth's polar motion.

stat.ME↗

arXiv2019

Identifying and Responding to Outlier Demand in Revenue Management

Revenue management strongly relies on accurate forecasts. Thus, when extraordinary events cause outlier demand, revenue management systems need to recognise this and adapt both forecast and controls. Many passenger transport service providers, such as railways and airlines, control the sale of tickets through revenue management. State-of-the-art systems in these industries rely on analyst expertise to identify outlier demand both online (within the booking horizon) and offline (in hindsight). So far, little research focuses on automating and evaluating the detection of outlier demand in this context. To remedy this, we propose a novel approach, which detects outliers using functional data analysis in combination with time series extrapolation. We evaluate the approach in a simulation framework, which generates outliers by varying the demand model. The results show that functional outlier detection yields better detection rates than alternative approaches for both online and offline analyses. Depending on the category of outliers, extrapolation further increases online detection performance. We also apply the procedure to a set of empirical data to demonstrate its practical implications. By evaluating the full feedback-driven system of forecast and optimisation, we generate insight on the asymmetric effects of positive and negative demand outliers. We show that identifying instances of outlier demand and adjusting the forecast in a timely fashion substantially increases revenue compared to what is earned when ignoring outliers.

stat.ME↗

arXiv2019

The Debiased Spatial Whittle Likelihood

We provide a computationally and statistically efficient method for estimating the parameters of a stochastic covariance model observed on a regular spatial grid in any number of dimensions. Our proposed method, which we call the Debiased Spatial Whittle likelihood, makes important corrections to the well-known Whittle likelihood to account for large sources of bias caused by boundary effects and aliasing. We generalise the approach to flexibly allow for significant volumes of missing data including those with lower-dimensional substructure, and for irregular sampling boundaries. We build a theoretical framework under relatively weak assumptions which ensures consistency and asymptotic normality in numerous practical settings including missing data and non-Gaussian processes. We also extend our consistency results to multivariate processes. We provide detailed implementation guidelines which ensure the estimation procedure can be conducted in O(n log n) operations, where n is the number of points of the encapsulating rectangular grid, thus keeping the computational scalability of Fourier and Whittle-based methods for large data sets. We validate our procedure over a range of simulated and real-world settings, and compare with state-of-the-art alternatives, demonstrating the enduring practical appeal of Fourier-based methods, provided they are corrected by the procedures developed in this paper.

stat.ME↗