arXiv subjects

Robin Mitra

Publications and source records attributed to Robin Mitra.

12 recordsLinked to original sources

arXiv2024

The appeal of the gamma family distribution to protect the confidentiality of contingency tables

Administrative databases, such as the English School Census (ESC), are rich sources of information that are potentially useful for researchers. For such data sources to be made available, however, strict guarantees of privacy would be required. To achieve this, synthetic data methods can be used. Such methods, when protecting the confidentiality of tabular data (contingency tables), often utilise the Poisson or Poisson-mixture distributions, such as the negative binomial (NBI). These distributions, however, are either equidispersed (in the case of the Poisson) or overdispersed (e.g. in the case of the NBI), which results in excessive noise being applied to large low-risk counts. This paper proposes the use of the (discretized) gamma family (GAF) distribution, which allows noise to be applied in a more bespoke fashion. Specifically, it allows less noise to be applied as cell counts become larger, providing an optimal balance in relation to the risk-utility trade-off. We illustrate the suitability of the GAF distribution on an administrative-type data set that is reminiscent of the ESC.

stat.ME↗

arXiv2024

How Deep is your Guess? A Fresh Perspective on Deep Learning for Medical Time-Series Imputation

We present a comprehensive analysis of deep learning approaches for Electronic Health Record (EHR) time-series imputation, examining how architectural and framework biases combine to influence model performance. Our investigation reveals varying capabilities of deep imputers in capturing complex spatiotemporal dependencies within EHRs, and that model effectiveness depends on how its combined biases align with medical time-series characteristics. Our experimental evaluation challenges common assumptions about model complexity, demonstrating that larger models do not necessarily improve performance. Rather, carefully designed architectures can better capture the complex patterns inherent in clinical data. The study highlights the need for imputation approaches that prioritise clinically meaningful data reconstruction over statistical accuracy. Our experiments show imputation performance variations of up to 20\% based on preprocessing and implementation choices, emphasising the need for standardised benchmarking methodologies. Finally, we identify critical gaps between current deep imputation methods and medical requirements, highlighting the importance of integrating clinical insights to achieve more reliable imputation approaches for healthcare applications.

cs.LG↗

arXiv2024

Obtaining $(\epsilon,\delta)$-differential privacy guarantees when using a Poisson mechanism to synthesize contingency tables

We show that differential privacy type guarantees can be obtained when using a Poisson synthesis mechanism to protect counts in contingency tables. Specifically, we show how to obtain $(\epsilon, \delta)$-probabilistic differential privacy guarantees via the Poisson distribution's cumulative distribution function. We demonstrate this empirically with the synthesis of an administrative-type confidential database.

cs.CR↗

arXiv2023

A Complete Characterisation of Structured Missingness

Our capacity to process large complex data sources is ever-increasing, providing us with new, important applied research questions to address, such as how to handle missing values in large-scale databases. Mitra et al. (2023) noted the phenomenon of Structured Missingness (SM), which is where missingness has an underlying structure. Existing taxonomies for defining missingness mechanisms typically assume that variables' missingness indicator vectors $M_1$, $M_2$, ..., $M_p$ are independent after conditioning on the relevant portion of the data matrix $\mathbf{X}$. As this is often unsuitable for characterising SM in multivariate settings, we introduce a taxonomy for SM, where each ${M}_j$ can depend on $\mathbf{M}_{-j}$ (i.e., all missingness indicator vectors except ${M}_j$), in addition to $\mathbf{X}$. We embed this new framework within the well-established decomposition of mechanisms into MCAR, MAR, and MNAR (Rubin, 1976), allowing us to recast mechanisms into a broader setting, where we can consider the combined effect of $\mathbf{X}$ and $\mathbf{M}_{-j}$ on ${M}_j$. We also demonstrate, via simulations, the impact of SM on inference and prediction, and consider contextual instances of SM arising in a de-identified nationwide (US-based) clinico-genomic database (CGDB). We hope to stimulate interest in SM, and encourage timely research into this phenomenon.

stat.ME↗

arXiv2023

Learning from data with structured missingness

Missing data are an unavoidable complication in many machine learning tasks. When data are `missing at random' there exist a range of tools and techniques to deal with the issue. However, as machine learning studies become more ambitious, and seek to learn from ever-larger volumes of heterogeneous data, an increasingly encountered problem arises in which missing values exhibit an association or structure, either explicitly or implicitly. Such `structured missingness' raises a range of challenges that have not yet been systematically addressed, and presents a fundamental hindrance to machine learning at scale. Here, we outline the current literature and propose a set of grand challenges in learning from data with structured missingness.

stat.ML↗

arXiv2022

A Framework for Auditable Synthetic Data Generation

Synthetic data has gained significant momentum thanks to sophisticated machine learning tools that enable the synthesis of high-dimensional datasets. However, many generation techniques do not give the data controller control over what statistical patterns are captured, leading to concerns over privacy protection. While synthetic records are not linked to a particular real-world individual, they can reveal information about users indirectly which may be unacceptable for data owners. There is thus a need to empirically verify the privacy of synthetic data -- a particularly challenging task in high-dimensional data. In this paper we present a general framework for synthetic data generation that gives data controllers full control over which statistical properties the synthetic data ought to preserve, what exact information loss is acceptable, and how to quantify it. The benefits of the approach are that (1) one can generate synthetic data that results in high utility for a given task, while (2) empirically validating that only statistics considered safe by the data curator are used to generate the data. We thus show the potential for synthetic data to be an effective means of releasing confidential data safely, while retaining useful information for analysts.

cs.CR↗

arXiv2022

An integrated approach to test for missing not at random

Missing data can lead to inefficiencies and biases in analyses, in particular when data are missing not at random (MNAR). It is thus vital to understand and correctly identify the missing data mechanism. Recovering missing values through a follow up sample allows researchers to conduct hypothesis tests for MNAR, which are not possible when using only the original incomplete data. Investigating how properties of these tests are affected by the follow up sample design is little explored in the literature. Our results provide comprehensive insight into the properties of one such test, based on the commonly used selection model framework. We determine conditions for recovery samples that allow the test to be applied appropriately and effectively, i.e. with known Type I error rates and optimized with respect to power. We thus provide an integrated framework for testing for the presence of MNAR and designing follow up samples in an efficient cost-effective way. The performance of our methodology is evaluated through simulation studies as well as on a real data sample.

stat.ME↗

arXiv2022

On integrating the number of synthetic data sets $m$ into the 'a priori' synthesis approach

Until recently, multiple synthetic data sets were always released to analysts, to allow valid inferences to be obtained. However, under certain conditions - including when saturated count models are used to synthesize categorical data - single imputation ($m=1$) is sufficient. Nevertheless, increasing $m$ causes utility to improve, but at the expense of higher risk, an example of the risk-utility trade-off. The question, therefore, is: which value of $m$ is optimal with respect to the risk-utility trade-off? Moreover, the paper considers two ways of analysing categorical data sets: as they have a contingency table representation, multiple categorical data sets can be averaged before being analysed, as opposed to the usual way of averaging post-analysis. This paper also introduces a pair of metrics, $\tau_3(k,d)$ and $\tau_4(k,d)$, that are suited for assessing disclosure risk in multiple categorical synthetic data sets. Finally, the synthesis methods are demonstrated empirically.

stat.ME↗

arXiv2021

Bayesian modelling and computation utilising directed cycles in multiple network data

Modelling multiple network data is crucial for addressing a wide range of applied research questions. However, there are many challenges, both theoretical and computational, to address. Network cycles are often of particular interest in many applications; for example in ecology a largely unexplored area has been how to incorporate network cycles within the inferential framework in an explicit way. The recently developed Spherical Network Family of models (SNF) offers a flexible formulation for modelling multiple network data that permits any type of metric. This has opened up the possibility to formulate network models that focus on network properties hitherto not possible or practical to consider. In this article we propose a novel network distance metric that measures similarities between networks with respect to their cycles, and incorporates this within the SNF model to allow inferences that explicitly capture information on cycles. These network motifs are of particular interest in ecological studies aimed at understanding competitive and hierarchical interactions. We further propose a novel computational framework to allow posterior inferences from the intractable SNF model for moderate-sized networks. Lastly, we apply the resulting methodology to a set of ecological network data studying aggressive interactions between species of fish. We show our model is able to make cogent inferences concerning the cycle behaviour amongst the species, and beyond those possible from a model that does not consider this network motif.

stat.AP↗

arXiv2021

Using saturated count models for user-friendly synthesis of categorical data

Over the past three decades, synthetic data methods for statistical disclosure control have continually evolved, but mainly within the domain of survey data sets. There are certain characteristics of administrative databases, such as their size, which present challenges from a synthesis perspective and require special attention. This paper, through the fitting of saturated count models, presents a synthesis method that is suitable for administrative databases that is tuned by two parameters. The method allows large categorical data sets to be synthesized quickly and allows risk and utility metrics to be satisfied a priori, that is, prior to synthetic data generation. The paper explores how the flexibility afforded by two-parameter count models (the negative binomial and Poisson-inverse Gaussian) can be utilised to protect respondents' - especially uniques' - privacy in synthetic data. Finally, an empirical example is carried out through the synthesis of a database which can be viewed as a good substitute to the English School Census.

stat.ME↗

arXiv2021

Bayesian model-based clustering for populations of network data

There is increasing appetite for analysing populations of network data due to the fast-growing body of applications demanding such methods. While methods exist to provide readily interpretable summaries of heterogeneous network populations, these are often descriptive or ad hoc, lacking any formal justification. In contrast, principled analysis methods often provide results difficult to relate back to the applied problem of interest. Motivated by two complementary applied examples, we develop a Bayesian framework to appropriately model complex heterogeneous network populations, whilst also allowing analysts to gain insights from the data, and make inferences most relevant to their needs. The first application involves a study in Computer Science measuring human movements across a University. The second analyses data from Neuroscience investigating relationships between different regions of the brain. While both applications entail analysis of a heterogeneous population of networks, network sizes vary considerably. We focus on the problem of clustering the elements of a network population, where each cluster is characterised by a network representative. We take advantage of the Bayesian machinery to simultaneously infer the cluster membership, the representatives, and the community structure of the representatives, thus allowing intuitive inferences to be made. The implementation of our method on the human movement study reveals interesting movement patterns of individuals in clusters, readily characterised by their network representative. For the brain networks application, our model reveals a cluster of individuals with different network properties of particular interest in Neuroscience. The performance of our method is additionally validated in extensive simulation studies.

stat.AP↗

arXiv2015

On the Use of Cauchy Prior Distributions for Bayesian Logistic Regression

In logistic regression, separation occurs when a linear combination of the predictors can perfectly classify part or all of the observations in the sample, and as a result, finite maximum likelihood estimates of the regression coefficients do not exist. Gelman et al. (2008) recommended independent Cauchy distributions as default priors for the regression coefficients in logistic regression, even in the case of separation, and reported posterior modes in their analyses. As the mean does not exist for the Cauchy prior, a natural question is whether the posterior means of the regression coefficients exist under separation. We prove theorems that provide necessary and sufficient conditions for the existence of posterior means under independent Cauchy priors for the logit link and a general family of link functions, including the probit link. We also study the existence of posterior means under multivariate Cauchy priors. For full Bayesian inference, we develop a Gibbs sampler based on Polya-Gamma data augmentation to sample from the posterior distribution under independent Student-t priors including Cauchy priors, and provide a companion R package in the supplement. We demonstrate empirically that even when the posterior means of the regression coefficients exist under separation, the magnitude of the posterior samples for Cauchy priors may be unusually large, and the corresponding Gibbs sampler shows extremely slow mixing. While alternative algorithms such as the No-U-Turn Sampler in Stan can greatly improve mixing, in order to resolve the issue of extremely heavy tailed posteriors for Cauchy priors under separation, one would need to consider lighter tailed priors such as normal priors or Student-t priors with degrees of freedom larger than one.

stat.ME↗