My Bachelor Thesis research from Maastricht Unviersity aiming to improve the accuracy of the acquisition processes at Q-Park. The thesis investigates different single- and multi-target regression algorithms and implements them in combination with conformal prediction framework that provides statistically valid guarantees on the results in form of prediction intervals.

My flagship project at Q-Park: an advanced application for simulating performance of new parking locations anywhere in Europe. The tool is implemented as a web application with Flask and uses Python as backend. The tool implements multiple methods and algorithms including data acquisition from many sources, k-NN comparison, classification, customer beharviour prediction and estimation of key performance indicators using multi-target conformal regression.

In this project for Q-Park, I used historical weather data and transformed them using a little-known, yet very powerful method called Weather Rating. Then comparing the weather rating for respective days with transaction volumes at car parks, I proved that there is a tring correlation between weather and parking behaviour and hence between weather and mobility.

A project I prepared for Q-Park with the goal of determining key revenue drivers of parking facilities based on the transaction patterns during COVID lockdowns and using k-means clustering. The method has successfully solved the given task and provided interesting insights into people's behaviour and how it changed during and right after the COVID lockdowns.

An article I wrote as a tutorial on how to plot API results and geo data on an interactive map in a web application. For the purpose of the article, I implemented a simple example, but I used the mapping feature in an advanced way when developping the Competitors Dashboard for Q-Park (see below). The solution I'm presenting uses Flask to create the web application and a LeafletJS library to create the map.

An advanced dashboard that I prepared for Q-Park to present data from Parkopedia API. It is built as a web application, using Flask library. The application uses Python for backend to acquire and process the data, and HTML, CSS and JavaScript for frontend (design, responsiveness, filtering). The dashboard is extensively used by multiple teams within Q-Park.

A collection of 4 projects I completed for a university course. The projects concern topics such as spacial filtering, geometric transformations, frequency filtering, Fourier Transform, noise removal, image restoration, data compression, morphological processing, binarization, and Motion Enegry Images (MEI).

A project for NGO of the European Youth Parliament Belgium ot improve its outreach to educational stakeholders. The goal was to create a database with contact information to all secondary schools in Belgium. The solution required a separate solution for each region in the country and, in some cases, web scraping using Selenium script.

Acquiring data about land use for many locations in Europe is something I used extensively when preparing models for Q-Park. One of the more important data sources was OpenStreetMap. The article I wrote explains how to count the number of buildings and places by category in various locations. The key issue I needed to solve were the inconsistencies in location tagging - the result of OpenStreetMap being a community-based project.

A short project, in which I was given a large set of addresses and a task to obtain a driving distance between all the locations. I used the Distance Matrix API from Google Maps, Given all limitations and caveats regarding this API, this task required a clever algorithm that turned out really interestingin the end.

A short project for an external client, in which I was given thousands of legal texts in form on HTML files. The first task was to extract the entire text from the HTML and format it as a CSV file. The second task involved finding definitions in the HTML files using a keyword and comparing how these definitions look in different files.