Side projects

intervalues is a Python and Rust package to efficiently combine intervals. What sets it apart from other similar packages is being able to track input values for these intervals, hence the name. Currently, there is a standalone Rust package, and a Rust-accelerated Python python package. In making these, I also created a new simplified Rust package to provide a very basic float approximation to circumvent some limitation with floats in Rust.

Counterstrike

One of my hobbies is watching competitive Counterstrike. For ranking CS teams, multiple rankings exist, provided by the maker of CS Valve, but also by tournament organizer ESL and by CS news and stats website HLTV. I have started a project to combine these, in 3 ways: (1) put them next to each other; (2) create simple aggregate statistics like average rank; and (3) producing an "optimal" combined ranking based on these. For this, I had to get access to the individual rankings as provided by the above parties, which has lead to the cs-rankings project (Github / PyPI). The project that uses this for the above 3 goals can be found here with this as the main output.

nl-maps

After moving to Bonaire, a special municipality of the Netherlands located in the Caribbean, I noticed that a lot of maps of the Netherlands don't show Bonaire (or the other two special municipalities, Saba and Sint Eustatius; the three together form the BES islands). Although in some cases this is defensible or even logical, in other cases it is simply an incomplete map, with election maps being the clearest example: for some of the elections like the national elections or the European elections, the votes on the BES islands has the same value per vote as the ones in the (European) Netherlands, so a map of "Most voted party per municipality" or "Voter turnout per municipality" without showing the BES islands is simply not complete. The data is often available to the news network (for example, in the interactive tool that the NOS has used for the past few elections you can select the three BES municipalities as well as postal voters from other countries). In order to promote the inclusion in future maps, I have created the package nl-maps (Github / PyPI) that can help with inserting the BES islands into existing Dutch informative maps. It can also be used to generate from scratch a map for the 12 provinces and the 3 BES islands, with support for each individual municipality planned for the future.

Personal projects

Some of my other personal projects I use for learning a new skill, maintaining a learned skill, or experimenting with new ideas can be found on my personal Github here. This includes stuff like my Advent of Code and Advent of SQL repositories, among others.

Language stats across these personal repositories

Research papers and publications

As a result of my PhD at the Tinbergen Institute and Erasmus University Rotterdam, I have been able to work on the following research papers and publications:

  • Managing Sales Forecasters, with Philip Hans Franses (DOI link), Tinbergen Institute Discussion Paper Series, December 2012.
  • Forecasting Earnings Forecasts, with Philip Hans Franses (Repub link), Tinbergen Institute Discussion Paper Series, January 2013.
  • Analyzing fixed-event forecast revisions, with Chia-Lin Chang, Philip Hans Franses, and Micheal McAleer (DOI link), International Journal of Forecasting, October 2013.
  • Stochastic levels and duration dependence in US unemployment, with Philip Hans Franses (Repub link), Erasmus University Rotterdam Econometric Institute Report Series, September 2015.
  • Heterogeneous Forecast Adjustment, with Philip Hans Franses (DOI link), Journal of Forecasting, July 2016.
  • Benchmarking judgmentally adjusted forecasts, with Philip Hans Franses (DOI link), International Journal of Finance and Economics, October 2016.
  • A novel approach to measuring consumer confidence, with Rene Segers and Philip Hans Franses (DOI link), Econometrics and Statistics, September 2017.

More details can be found by following the mentioned DOI or Repub links. Most of my research back then was done using Matlab, EViews, Python, and/or LaTeX.