Research

Most data about people cannot be handed on as they are, and most datasets have holes. My research asks how data can be shared, completed and checked. Five themes, each with five selected papers and the packages behind it; the full list is on the publications page.

Anonymisation and synthetic data

Statistical disclosure control measures how likely it is that someone in a dataset can be recognised and changes the data just enough to make that unlikely while keeping analyses valid. Synthetic data go one step further and replace every record by a plausible one. I work on risk measures beyond k-anonymity, on longitudinal and mobility data, on risk-utility maps for choosing a method, and on what agentic AI changes for attackers and defenders.

Packages: sdcMicro, simPop, riskutility, synvey, robSynth

  1. 2026Thees, Müller, Templ. Agentic AI-Powered Re-identification: An Emerging, Scalable Threat to Mobility Microdata Privacy. Privacy in Statistical Databases. PSD 2026 doi
  2. 2024Gussenbauer, Templ, Fritzmann, Kowarik. Simulation of Calibrated Complex Synthetic Population Data with XGBoost. Algorithms doi
  3. 2022Templ, Sariyar. A systematic overview on methods to protect sensitive data provided for various analyses. International Journal of Information Security doi
  4. 2017Templ, Meindl, Kowarik, Dupriez. Simulation of Synthetic Complex Data: The R Package simPop. Journal of Statistical Software doi
  5. 2015Templ, Kowarik, Meindl. Statistical Disclosure Control for Micro-Data Using the R Package sdcMicro. Journal of Statistical Software doi

Missing values and imputation

Before values can be imputed one has to see where and why they are missing. VIM visualises missingness and offers imputation methods that are robust to outliers; more recent work uses generalised additive models and deep learning and studies what misclassifications and outliers do to imputation.

Packages: VIM, deepImp

  1. 2024Templ. Robust multiple imputation with GAM. Statistics and Computing doi
  2. 2024Templ, Ulmer. The impact of misclassifications and outliers on imputation methods. Journal of Applied Statistics doi
  3. 2023Templ. Visualization and Imputation of Missing Values. Springer Publishing doi
  4. 2016Kowarik, Templ. Imputation with the R Package VIM. Journal of Statistical Software doi
  5. 2012Templ, Alfons, Filzmoser. Exploring incomplete data using visualization techniques. Advances in Data Analysis and Classification

Compositional data

Shares that sum to one hundred percent, geochemical concentrations, budgets, time-use diaries: compositional data live on a simplex, and standard methods go wrong there. With Peter Filzmoser and Karel Hron I work on robust methods, imputation of missing and rounded-zero parts and, more recently, on densities as compositions in Bayes spaces.

Packages: robCompositions

  1. 2018Filzmoser, Hron, Templ. Applied Compositional Data Analysis. With Worked Examples in R. Springer Series in Statistics doi
  2. 2015Martín-Fernández, Hron, Templ, Filzmoser, Palarea-Albaladejo. Bayesian-multiplicative treatment of count zeros in compositional data sets. Statistical Modelling link
  3. 2012Martín-Fernández, Hron, Templ, Filzmoser, Palarea-Albaladejo. Model-based replacement of rounded zeros in compositional data: Classical and robust approaches. Computational Statistics and Data Analysis
  4. 2011Templ, Hron, Filzmoser. robCompositions: an R-package for robust statistical analysis of compositional data. Compositional Data Analysis: Theory and Applications doi
  5. 2010Hron, Templ, Filzmoser. Imputation of missing values for compositional data using classical and robust methods. Computational Statistics and Data Analysis

Robust statistics

Real data contain outliers, and in modern data the contamination sits in single cells rather than whole rows. I work on cellwise-robust imputation and synthetic data, on robust estimation of indicators from complex surveys and on outlier detection that scales.

Packages: robCompositions, robSynth, laeken

  1. 2019Templ, Gussenbauer, Filzmoser. Evaluation of robust outlier detection methods for zero-inflated complex data. Journal of Applied Statistics doi
  2. 2013Alfons, Templ, Filzmoser. Robust estimation of economic indicators from survey samples based on Pareto tail modelling. Journal of the Royal Statistical Society: Series C (Applied Statistics) doi
  3. 2011Templ, Kowarik, Filzmoser. Iterative stepwise regression imputation using standard and robust methods. Computational Statistics and Data Analysis
  4. 2011Todorov, Templ, Filzmoser. Detection of multivariate outliers in business survey data with incomplete information. Advances in Data Analysis and Classification
  5. 2008Templ, Meindl. Robust Statistics Meets SDC: New Disclosure Risk Measures for Continuous Microdata Masking. Lecture Notes on Computer Science doi

Official and survey statistics

Twelve years at Statistics Austria shaped how I think about data quality: sampling, weighting, editing, imputation and release are one pipeline. I co-maintain the CRAN task view on official statistics and co-wrote a German textbook on data quality in sample surveys.

Packages: laeken, emdi, simPop, heaping

  1. 2026Quatember, Templ. Datenqualität in Stichprobenerhebungen: Eine verständnisorientierte Einführung in die Survey-Statistik. Springer Spektrum doi
  2. 2019Kreutzmann, Pannier, Rojas-Perilla, Schmid, Templ, Tzavidis. The R Package emdi for the Estimation and Mapping of Regional Disaggregated Indicators. Journal of Statistical Software doi
  3. 2015Templ. Quality Indicators for Statistical Disclosure Methods: A Case Study on the Structure of Earnings Survey. Journal of Official Statistics doi
  4. 2013Alfons, Templ. Estimation of Social Exclusion Indicators from Complex Surveys: The R Package laeken. Journal of Statistical Software doi
  5. 2010Alfons, Templ, Filzmoser, Holzer, González-Rodríguez. A comparison of robust methods for Pareto tail modeling in the case of Laeken indicators. Combining Soft Computing and Statistical Methods in Data Analysis

Center for Data Quality and Anonymization

At the FHNW School of Business the center (C4DataQuality) bundles data-quality analysis, survey-statistics consulting, strategies for missing data, anonymisation and data release, and training. Team: Tobias Schoch, Roman Müller, Oscar Thees.

Page at FHNW

SwissAnon

The Swiss Data Anonymization Center, anchored at FHNW and run with Murat Sariyar (BFH), with UZH, BFH and ZHAW as partners. Research, courses, software and consulting for companies and public bodies.

swissanon.com