MSc in Business Intelligence
Focused on analytics, automation, and creating impactful data solutions.
Skilled in Power BI, Python, R, SQL, Excel, and Docker
I design machine learning models, interactive dashboards, and
automated pipelines that turn complex data into smart, scalable solutions
from predicting customer behavior to uncovering deep market insights.
This project showcases an interactive dashboard designed to predict customer churn using a trained XGBoost model and SHAP-based feature explanations. With intuitive sliders, CSV upload, and real-time insights, the tool empowers businesses to identify at-risk customers, understand the drivers behind churn, and take proactive retention measures.
A responsive personal portfolio website designed to present academic background, technical skills, and data-related projects. Built using HTML, CSS, and the HTML5UP template, this project is fully hosted on GitHub Pages. It features dynamic sections like resume downloads, project showcases, and a contact form integrated with Formspree for direct messaging.
This project leverages real-time API data, weather conditions, and traffic congestion to forecast parking availability across Aarhus. Using machine learning models like Random Forest and XGBoost, it predicts occupancy levels and visualizes trends through interactive dashboards—helping drivers find parking faster and reducing city congestion.
This project focuses on building a classification model to predict mobile phone price ranges based on their technical specifications. Using a dataset of 2,000 phones with 21 features, the model helps identify which features influence pricing the most, supporting strategic decisions for market positioning and product segmentation.
This project applies convolutional neural networks (CNNs) to candlestick chart images to predict whether a stock’s price will go up or down the next day. A convolutional neural network that classifies candlestick chart images to predict next-day stock price direction. The model is trained from scratch and optimized through learning rate tuning for improved accuracy on test data.