Open resources
Resources
Open Science Study Group
An informal space for open and reproducible research. The study group provides a friendly platform for discussions about open science, reproducibility, and meta-science. Its aims include helping students and researchers become familiar with open-science practices, connecting people with shared interests, and promoting critical discussion of research methods.
Before each meeting, participants read a selected article and discuss its implications for research practice during the session.
See ReproducibiliTea
Psychometrics
ScaleCheck: Psychometric Assumption Diagnostics for Reflective Measurement
ScaleCheck is an R Shiny app to assess measurement assumptions for reflective constructs. It integrates unidimensionality, latent monotonicity, and local dependence checks in a single workflow. The app provides a dynamic interpretation to aid user understand the results.
Controlling Social Desirability: quadSim and quadSimple
I designed computational tools for simulation-based assessment of statistical power, parameter bias, and confidence-interval coverage in structural equation models using the Multiple Indicators Multiple Causes (MIMIC) Quadruplets approach.
quadSimple offers an accessible workflow for early instrument-development decisions, while quadSim supports more detailed simulations when researchers already have information about the instrument parameters.
quadSimple – quadSim – Source Code
Machine-learning projects
End-to-end projects covering data preparation, feature engineering, predictive modeling, evaluation, interpretation, and deployment.
Forecasting Magic: The Gathering Card Prices
A machine-learning project predicting the market value of Magic: The Gathering cards from characteristics such as rarity, type, release set, and other structured features.
The workflow includes data collection and cleaning, exploratory analysis, feature engineering, regression modeling, performance evaluation, and interpretation of the features associated with price variation.
Hair Condition Classification from Image Data
An end-to-end image-classification pipeline designed to distinguish alopecia, receding hairline, and no alopecia using Histogram of Oriented Gradients features and a support vector machine.
The project includes standardized preprocessing, class balancing, multiclass evaluation, confidence scores, and an interactive Streamlit deployment for real-time predictions.
Open the Application – View Notebook and Model