arXiv subjects

Ioanna Manolopoulou

Publications and source records attributed to Ioanna Manolopoulou.

At least 19 recordsLinked to original sources

arXiv2026

Improving exoplanet mass characterisation with Bayesian model selection using the Learned Harmonic Mean Estimator

Radial velocity (RV) analyses require modelling choices (such as eccentricity treatment, noise model, velocity trends, and number of planets) that can significantly affect derived planetary masses. Current practice often relies on information criteria to compare and select models, but these have known limitations: they lack the built-in Occam's razor of Bayesian model comparison, and they do not incorporate prior information. Computing the Bayesian evidence needed for Bayes factor model comparison has traditionally required dedicated algorithms such as nested sampling. The learned harmonic mean estimator (LHME) offers an alternative, estimating the Bayesian evidence directly from MCMC posterior samples, with less computational cost and with no modification to the fitting procedure. We present the first application of the LHME to RV model selection, fitting 18 model variants -- comparing circular and eccentric orbits, white noise and Gaussian Process noise models, and long-term velocity trends -- to six single-planet systems, and 72 variants to a seventh system for an $N$ versus $N+1$ planet model comparison. We find that no single model is universally preferred, reinforcing the need for model comparison to select the most appropriate model for a system, thereby ensuring robust mass characterisation. The LHME, implemented in the open-source harmonic package, makes rigorous Bayesian model comparison accessible to existing MCMC-based RV workflows, and we encourage its wider use for other model comparisons in astrophysics.

astro-ph.EP↗

arXiv2025

A real-time metric of online engagement monitoring

Measuring online behavioural student engagement often relies on simple count indicators or retrospective, predictive methods, which present challenges for real-time application. To address these limitations, we reconceptualise an existing course-wide engagement metric to create a chapter-based version that aligns with the weekly structure of online courses. Derived directly from virtual learning environment log data, the new metric allows for cumulative, real-time tracking of student activity without requiring outcome data or model training. We evaluate the approach across three undergraduate statistics modules over two academic years, comparing it to the course-wide formulation to assess how the reconceptualisation influences what is measured. Results indicate strong alignment from as early as week 3, along with comparable or improved predictive validity for final grades in structured, lecture-based contexts. By the course midpoint, the weekly metric identifies as many low-performing students as are identifiable by the end of the course. While performance varies across modules, the chapter-based formulation offers a scalable and interpretable method for early engagement monitoring and student support.

cs.CY↗

arXiv2025

Bayesian Nonparametric Models for Multiple Raters: a General Statistical Framework

Rating procedure is crucial in many applied fields (e.g., educational, clinical, emergency). It implies that a rater (e.g., teacher, doctor) rates a subject (e.g., student, doctor) on a rating scale. Given raters variability, several statistical methods have been proposed for assessing and improving the quality of ratings. Model estimation in the presence of heterogeneity has been one of the recent challenges in this research line. Consequently, several methods have been proposed to address this issue under a parametric multilevel modelling framework, in which strong distributional assumptions are made. We propose a more flexible model under the Bayesian nonparametric (BNP) framework, in which most of those assumptions are relaxed. By eliciting hierarchical discrete nonparametric priors, the model accommodates clusters among raters and subjects, naturally accounts for heterogeneity, and improves estimates accuracy. We propose a general BNP heteroscedastic framework to analyse continuous and coarse rating data and possible latent differences among subjects and raters. The estimated densities are used to make inferences about the rating process and the quality of the ratings. By exploiting a stick-breaking representation of the Dirichlet Process, a general class of Intraclass Correlation Coefficient (ICC) indices might be derived for these models. Our method allows us to independently identify latent similarities between subjects and raters and can be applied in precise education to improve personalised teaching programs or interventions. Theoretical results about the ICC are provided together with computational strategies. Simulations and a real-world application are presented, and possible future directions are discussed.

stat.ME↗

arXiv2024

Uncovering Student Engagement Patterns in Moodle with Interpretable Machine Learning

Understanding and enhancing student engagement through digital platforms is critical in higher education. This study introduces a methodology for quantifying engagement across an entire module using virtual learning environment (VLE) activity log data. Using study session frequency, immediacy, and diversity, we create a cumulative engagement metric and model it against weekly VLE interactions with resources to identify critical periods and resources predictive of student engagement. In a case study of a computing module at University College London's Department of Statistical Science, we further examine how delivery methods (online, hybrid, in-person) impact student behaviour. Across nine regression models, we validate the consistency of the random forest model and highlight the interpretive strengths of generalised additive models for analysing engagement patterns. Results show weekly VLE clicks as reliable engagement predictors, with early weeks and the first assessment period being key. However, the impact of delivery methods on engagement is inconclusive due to inconsistencies across models. These findings support early intervention strategies to assist students at risk of disengagement. This work contributes to learning analytics research by proposing a refined VLE-based engagement metric and advancing data-driven teaching strategies in higher education.

cs.CY↗

arXiv2023

Mixture polarization in inter-rater agreement analysis: a Bayesian nonparametric index

In several observational contexts where different raters evaluate a set of items, it is common to assume that all raters draw their scores from the same underlying distribution. However, a plenty of scientific works have evidenced the relevance of individual variability in different type of rating tasks. To address this issue the intra-class correlation coefficient (ICC) has been used as a measure of variability among raters within the Hierarchical Linear Models approach. A common distributional assumption in this setting is to specify hierarchical effects as independent and identically distributed from a normal with the mean parameter fixed to zero and unknown variance. The present work aims to overcome this strong assumption in the inter-rater agreement estimation by placing a Dirichlet Process Mixture over the hierarchical effects' prior distribution. A new nonparametric index $λ$ is proposed to quantify raters polarization in presence of group heterogeneity. The model is applied on a set of simulated experiments and real world data. Possible future directions are discussed.

stat.ME↗

arXiv2022

Counterfactual Learning with Multioutput Deep Kernels

In this paper, we address the challenge of performing counterfactual inference with observational data via Bayesian nonparametric regression adjustment, with a focus on high-dimensional settings featuring multiple actions and multiple correlated outcomes. We present a general class of counterfactual multi-task deep kernels models that estimate causal effects and learn policies proficiently thanks to their sample efficiency gains, while scaling well with high dimensions. In the first part of the work, we rely on Structural Causal Models (SCM) to formally introduce the setup and the problem of identifying counterfactual quantities under observed confounding. We then discuss the benefits of tackling the task of causal effects estimation via stacked coregionalized Gaussian Processes and Deep Kernels. Finally, we demonstrate the use of the proposed methods on simulated experiments that span individual causal effects estimation, off-policy evaluation and optimization.

cs.LG↗

arXiv2022

Interpretable Deep Causal Learning for Moderation Effects

In this extended abstract paper, we address the problem of interpretability and targeted regularization in causal machine learning models. In particular, we focus on the problem of estimating individual causal/treatment effects under observed confounders, which can be controlled for and moderate the effect of the treatment on the outcome of interest. Black-box ML models adjusted for the causal setting perform generally well in this task, but they lack interpretable output identifying the main drivers of treatment heterogeneity and their functional relationship. We propose a novel deep counterfactual learning architecture for estimating individual treatment effects that can simultaneously: i) convey targeted regularization on, and produce quantify uncertainty around the quantity of interest (i.e., the Conditional Average Treatment Effect); ii) disentangle baseline prognostic and moderating effects of the covariates and output interpretable score functions describing their relationship with the outcome. Finally, we demonstrate the use of the method via a simple simulated experiment.

cs.LG↗

arXiv2021

Estimating Individual Treatment Effects using Non-Parametric Regression Models: a Review

Large observational data are increasingly available in disciplines such as health, economic and social sciences, where researchers are interested in causal questions rather than prediction. In this paper, we examine the problem of estimating heterogeneous treatment effects using non-parametric regression-based methods, starting from an empirical study aimed at investigating the effect of participation in school meal programs on health indicators. Firstly, we introduce the setup and the issues related to conducting causal inference with observational or non-fully randomized data, and how these issues can be tackled with the help of statistical learning tools. Then, we review and develop a unifying taxonomy of the existing state-of-the-art frameworks that allow for individual treatment effects estimation via non-parametric regression models. After presenting a brief overview on the problem of model selection, we illustrate the performance of some of the methods on three different simulated studies. We conclude by demonstrating the use of some of the methods on an empirical analysis of the school meal program data.

stat.ME↗

arXiv2021

Shrinkage Bayesian Causal Forests for Heterogeneous Treatment Effects Estimation

This paper develops a sparsity-inducing version of Bayesian Causal Forests, a recently proposed nonparametric causal regression model that employs Bayesian Additive Regression Trees and is specifically designed to estimate heterogeneous treatment effects using observational data. The sparsity-inducing component we introduce is motivated by empirical studies where not all the available covariates are relevant, leading to different degrees of sparsity underlying the surfaces of interest in the estimation of individual treatment effects. The extended version presented in this work, which we name Shrinkage Bayesian Causal Forest, is equipped with an additional pair of priors allowing the model to adjust the weight of each covariate through the corresponding number of splits in the tree ensemble. These priors improve the model's adaptability to sparse data generating processes and allow to perform fully Bayesian feature shrinkage in a framework for treatment effects estimation, and thus to uncover the moderating factors driving heterogeneity. In addition, the method allows prior knowledge about the relevant confounding covariates and the relative magnitude of their impact on the outcome to be incorporated in the model. We illustrate the performance of our method in simulated studies, in comparison to Bayesian Causal Forest and other state-of-the-art models, to demonstrate how it scales up with an increasing number of covariates and how it handles strongly confounded scenarios. Finally, we also provide an example of application using real-world data.

stat.ME↗

arXiv2021

Regional Topics in British Grocery Retail Transactions

Understanding the customer behaviours behind transactional data has high commercial value in the grocery retail industry. Customers generate millions of transactions every day, choosing and buying products to satisfy specific shopping needs. Product availability may vary geographically due to local demand and local supply, thus driving the importance of analysing transactions within their corresponding store and regional context. Topic models provide a powerful tool in the analysis of transactional data, identifying topics that display frequently-bought-together products and summarising transactions as mixtures of topics. We use the Segmented Topic Model (STM) to capture customer behaviours that are nested within stores. STM not only provides topics and transaction summaries but also topical summaries at the store level that can be used to identify regional topics. We summarised the posterior distribution of STM by post-processing multiple posterior samples and selecting semantic modes represented as recurrent topics. We use linear Gaussian process regression to model topic prevalence across British territory while accounting for spatial autocorrelation. We implement our methods on a dataset of transactional data from a major UK grocery retailer and demonstrate that shopping behaviours may vary regionally and nearby stores tend to exhibit similar regional demand.

stat.AP↗

arXiv2021

Modelling Grocery Retail Topic Distributions: Evaluation, Interpretability and Stability

Understanding the shopping motivations behind market baskets has high commercial value in the grocery retail industry. Analyzing shopping transactions demands techniques that can cope with the volume and dimensionality of grocery transactional data while keeping interpretable outcomes. Latent Dirichlet Allocation (LDA) provides a suitable framework to process grocery transactions and to discover a broad representation of customers' shopping motivations. However, summarizing the posterior distribution of an LDA model is challenging, while individual LDA draws may not be coherent and cannot capture topic uncertainty. Moreover, the evaluation of LDA models is dominated by model-fit measures which may not adequately capture the qualitative aspects such as interpretability and stability of topics. In this paper, we introduce clustering methodology that post-processes posterior LDA draws to summarise the entire posterior distribution and identify semantic modes represented as recurrent topics. Our approach is an alternative to standard label-switching techniques and provides a single posterior summary set of topics, as well as associated measures of uncertainty. Furthermore, we establish a more holistic definition for model evaluation, which assesses topic models based not only on their likelihood but also on their coherence, distinctiveness and stability. By means of a survey, we set thresholds for the interpretation of topic coherence and topic similarity in the domain of grocery retail data. We demonstrate that the selection of recurrent topics through our clustering methodology not only improves model likelihood but also outperforms the qualitative aspects of LDA such as interpretability and stability. We illustrate our methods on an example from a large UK supermarket chain.

stat.AP↗

arXiv2019

Incorporating Astrochemistry into Molecular Line Modelling via Emulation

In studies of the interstellar medium in galaxies, radiative transfer models of molecular emission are useful for relating molecular line observations back to the physical conditions of the gas they trace. However, doing this requires solving a highly degenerate inverse problem. In order to alleviate these degeneracies, the abundances derived from astrochemical models can be converted into column densities and fed into radiative transfer models. This enforces that the molecular gas composition used by the radiative transfer models be chemically realistic. However, because of the complexity and long running time of astrochemical models, it can be difficult to incorporate chemical models into the radiative transfer framework. In this paper, we introduce a statistical emulator of the UCLCHEM astrochemical model, built using neural networks. We then illustrate, through examples of parameter estimations, how such an emulator can be applied on real and synthetic observations.

astro-ph.GA↗

arXiv2018

BPEC: An R Package for Bayesian Phylogeographic and Ecological Clustering

BPEC is an R package for Bayesian Phylogeographic and Ecological Clustering which allows geographical, environmental and phenotypic measurements to be combined with DNA sequences in order to reveal clustered structure resulting from migration events. DNA sequences are modelled using a collapsed version of a simplified coalescent model projected onto haplotype trees, which subsequently give rise to constrained clusterings as migrations occur. Within each cluster, a multivariate Gaussian distribution of the covariates (geographical, environmental, phenotypic) is used. Inference follows tailored Reversible Jump Markov chain Monte Carlo sampling so that the number of clusters (i.e., migrations) does not need to be pre-specified. A number of output plots and visualizations are provided which reflect the posterior distribution of the parameters of interest. BPEC also includes functions that create output files which can be loaded into Google Earth. The package commands are illustrated through an example dataset of the polytypic Near Eastern brown frog Rana macrocnemis analysed using BPEC.

stat.AP↗

arXiv2018

Dirichlet Process Mixtures of Order Statistics with Applications to Retail Analytics

The rise of "big data" has led to the frequent need to process and store datasets containing large numbers of high dimensional observations. Due to storage restrictions, these observations might be recorded in a lossy-but-sparse manner, with information collapsed onto a few entries which are considered important. This results in informative missingness in the observed data. Our motivating application comes from retail analytics, where the behaviour of product sales is summarised by the price elasticity of each product with respect to a small number of its top competitors. The resulting data are vectors of order statistics, due to only the top few entries being observed. Interest lies in characterising the behaviour of a product's competitors, and clustering products based on how their competition is spread across the market. We develop nonparametric Bayesian methodology for modelling vectors of order statistics that utilises a Dirichlet Process Mixture Model with an Exponentiated Weibull kernel. Our approach allows us added flexibility for the distribution of each vector, while providing parameters that characterise the decay of the leading entries. We implement our methods on a retail analytics dataset of the cross-elasticity coefficients, and our analysis reveals distinct types of behaviour across the different products of interest.

stat.AP↗

arXiv2018

Bayesian hierarchical modelling of sparse count processes in retail analytics

The field of retail analytics has been transformed by the availability of rich data which can be used to perform tasks such as demand forecasting and inventory management. However, one task which has proved more challenging is the forecasting of demand for products which exhibit very few sales. The sparsity of the resulting data limits the degree to which traditional analytics can be deployed. To combat this, we represent sales data as a structured sparse multivariate point process which allows for features such as auto-correlation, cross-correlation, and temporal clustering, known to be present in sparse sales data. We introduce a Bayesian point process model to capture these phenomena, which includes a hurdle component to cope with sparsity and an exciting component to cope with temporal clustering within and across products. We then cast this model within a Bayesian hierarchical framework, to allow the borrowing of information across different products, which is key in addressing the data sparsity per product. We conduct a detailed analysis using real sales data to show that this model outperforms existing methods in terms of predictive power and we discuss the interpretation of the inference.

stat.AP↗

arXiv2018

Estimating the Expected Value of Sample Information across Different Sample Sizes using Moment Matching and Non-Linear Regression

Background: The Expected Value of Sample Information (EVSI) determines the economic value of any future study with a specific design aimed at reducing uncertainty in a health economic model. This has potential as a tool for trial design; the cost and value of different designs could be compared to find the trial with the greatest net benefit. However, despite recent developments, EVSI analysis can be slow especially when optimising over a large number of different designs. Methods: This paper develops a method to reduce the computation time required to calculate the EVSI across different sample sizes. Our method extends the moment matching approach to EVSI estimation to optimise over different sample sizes for the underlying trial with a similar computational cost to a single EVSI estimate. This extension calculates posterior variances across the alternative sample sizes and then uses Bayesian non-linear regression to calculate the EVSI. Results: A health economic model developed to assess the cost-effectiveness of interventions for chronic pain demonstrates that this EVSI calculation method is fast and accurate for realistic models. This example also highlights how different trial designs can be compared using the EVSI. Conclusion: The proposed estimation method is fast and accurate when calculating the EVSI across different sample sizes. This will allow researchers to realise the potential of using the EVSI to determine an economically optimal trial design for reducing uncertainty in health economic models. Limitations: Our method relies on some additional simulation, which can be expensive in models with very large computational cost.

stat.ME↗

arXiv2017

Efficient Monte Carlo Estimation of the Expected Value of Sample Information using Moment Matching

The Expected Value of Sample Information (EVSI) is used to calculate the economic value of a new research strategy. While this value would be important to both researchers and funders, there are very few practical applications of the EVSI. In the main, this is due to computational difficulties associated with calculating the EVSI in practical health economic models using nested simulations. We present an approximation method for the EVSI that is based on estimating the distribution of the posterior mean of the incremental net benefit across all the possible future samples, known as the distribution of the preposterior mean. Specifically, we suggest that this distribution is estimated using moment matching coupled with simulations that are available for probabilistic sensitivity analysis, which is typically mandatory in health economic evaluation. We demonstrate that this method is successful using an example that has previously been applied to other EVSI approximation methods. We then conclude by discussing how our method fits in with other recent additions to the literature that detail approximation methods for the EVSI.

math.ST↗

arXiv2016

Estimating the Expected Value of Partial Perfect Information in Health Economic Evaluations using Integrated Nested Laplace Approximation

The Expected Value of Perfect Partial Information (EVPPI) is a decision-theoretic measure of the "cost" of parametric uncertainty in decision making used principally in health economic decision making. Despite this decision-theoretic grounding, the uptake of EVPPI calculations in practice has been slow. This is in part due to the prohibitive computational time required to estimate the EVPPI via Monte Carlo simulations. However, recent developments have demonstrated that the EVPPI can be estimated by non-parametric regression methods, which have significantly decreased the computation time required to approximate the EVPPI. Under certain circumstances, high-dimensional Gaussian Process regression is suggested, but this can still be prohibitively expensive. Applying fast computation methods developed in spatial statistics using Integrated Nested Laplace Approximations (INLA) and projecting from a high-dimensional into a low-dimensional input space allows us to decrease the computation time for fitting these high-dimensional Gaussian Processes, often substantially. We demonstrate that the EVPPI calculated using our method for Gaussian Process regression is in line with the standard Gaussian Process regression method and that despite the apparent methodological complexity of this new method, R functions are available in the package BCEA to implement it simply and efficiently.

stat.AP↗