Ondrej Hoberla - Portfolio

Research

Covid-19 Vaccination Uptake in the Czech Republic

Data Management & Visualisastion

Skills employed: programming (R), data science (data extraction, manipulation, and transformation), Data visualisation and analytics (data exploration, visualisation selection and customisation) Areas of research: public health, health statistics

The aim of this project was to visualise Covid-19 vaccine uptake in the Czech Republic. Completing this project required engaging with the Czech healthcare system, extracting data from public healthcare repositories, cleaning the data, and matching it with public statistical reports to provide as accurate region-based estimates as possible. The project site depicts the step-by-step process of creating the visualisation, which is accompanied by a critical appraisal and interpretation.

Examining a Covariate Multiverse informed by Crowdsourced and Multiverse Analyses of the ‘Many Analysts, One Dataset’ Study

Data Science & Advanced Statistics

Skills employed: R programming, Python programming, Data science - data extraction, Advanced statistics - multilevel modelling, logistic regression, critical research evaluation
Areas of research: open Science, sports statistics, social psychology, cognitive psychology

Making decisions in research leads to researcher degrees of freedom that can be seen as exploring a subset of options and outcomes. They lead to over-reporting significance and under-reporting non-significance. Globally, this trend contributed to the replication crisis in psychology; a large proportion of reported effects cannot be replicated because they are unlikely to exist. Principles of Open science such as pre-registration, transparency, full procedure disclosure and data sharing aim to eliminate reporting biases but are faced with practical limitations and the issue of parallel ‘viable’ choices leading to different results. Crowdsourced studies further identified idiosyncratic variability unaccounted for by experience, knowledge, procedures, or peer-rated study quality. Multiverse analyses allow researchers to analyse all possible outcomes of any given specification, which is researcher-driven and therefore cannot eliminate human choices. Multiverse analyses of a sports data set are discussed whereby the impact of covariates and functional forms on whether skin tone ratings can predict the award of red cards is examined. This study provides new specifications to extend the multiverse and explores related issues of overall multiverse model performance and average model performance based on covariate grouping.

Cleaned data capturing player-referee interactions are used to construct a multiverse of multilevel logistic regression models with red card awards an outcome, averaged skin tone ratings a predictor, and a selection of covariates. Non-independence is treated using random intercepts.

Overfitting was detected and eliminated. Model performance was modest both overall and after grouping models by included covariates. Estimates of skin tone ratings showed a stronger effect and a high proportion of statistical significance after overfitting elimination. This revealed a change in estimate distribution compared to previous studies.

Multiverse analyses have excellent exploratory power but are limited by human choices. They should be used in conjunction with replication or crowdsourcing. Limitations and future directions are discussed.

Data Analytics Portfolio

Captsone Project (Google Data Analytics Certification)

Coming soon

Get in touch!

Personal website / Linkedin profile / GitHub Profile

About me

Enthusiastic Teaching Fellow with 8 years of diverse work experience across roles focused on data processing and analytics. Utilises a strong background in quantitative research skills and applied statistics while demonstrating flexibility, responsiveness, and ability to learn quickly across the varied past roles. Expertise in project administration and stakeholder-centred approach to problem solving enhances the ability to carefully examine contexts, analyse data, and extract meaningful insights. Commitment to continuous development and innovation enhanced by working in collaborative environments and a desire to demonstrate professional excellence in contributing to meaningful work with strong positive impact provides excellent foundations to embrace challenges in the rapidly evolving field of data analytics and science.

Education

Achievements & Grants

Recent employment

Skills and certifications