Daniel Hellmuth

> Built a machine learning pipeline (SQL, Python) to identify elite strikers using Opta data, combining UMAP for dimension reduction and HDBSCAN clustering, achieving a silhouette score of 0.62 across 16 clusters.

> Identified high-performing, cost-efficient forwards (e.g. Mateo Retegui) from 216 players in Europe’s top leagues; grouped targets by scoring efficiency on a per 90 basis.

> Delivered insights through data storytelling, visualisation (Matplotlib, Seaborn), and… Show more

> Built a machine learning pipeline (SQL, Python) to identify elite strikers using Opta data, combining UMAP for dimension reduction and HDBSCAN clustering, achieving a silhouette score of 0.62 across 16 clusters.

> Identified high-performing, cost-efficient forwards (e.g. Mateo Retegui) from 216 players in Europe’s top leagues; grouped targets by scoring efficiency on a per 90 basis.

> Delivered insights through data storytelling, visualisation (Matplotlib, Seaborn), and a GitHub portfolio project, demonstrating end-to-end analytical thinking and football intelligence. Show less

> Designed and implemented a forecasting pipeline in Python to predict daily item sales across Ecuadorian stores, reducing baseline MAPE from 8.09% to 1.66% using tuned XGBoost.

> Independently cleaned, joined, and engineered features from complex datasets (sales, promotions, oil prices, holidays), applying time series best practices and rolling validation.

> Delivered clear visualizations and documentation, demonstrating analytical thinking, self-motivation, and… Show more

> Designed and implemented a forecasting pipeline in Python to predict daily item sales across Ecuadorian stores, reducing baseline MAPE from 8.09% to 1.66% using tuned XGBoost.

> Independently cleaned, joined, and engineered features from complex datasets (sales, promotions, oil prices, holidays), applying time series best practices and rolling validation.

> Delivered clear visualizations and documentation, demonstrating analytical thinking, self-motivation, and effective communication of technical findings. Show less

> Filtered (SQL) and clustered 6,730 loyal customers in Python using K-Means and Hierarchical methods to uncover 5 actionable segments (e.g. Business Traveller, Family Booker).

> Mapped each segment to a loyalty perk using behavioural insights (e.g. booking frequency, discount sensitivity, travel length).

> Delivered findings through stakeholder friendly charts (Matplotlib, Seaborn), GoogleSlides and an executive summary.

> Analyzed 11,000+ real-world car records to uncover patterns, trends and anomalys between features using Pandas, NumPy, Matplotlib and Seaborn in Google Colab.

> Cleaned and engineered features; found strong correlation between engine HP and MSRP (0.65) and identified high-price outliers like Bugatti.

> Managed the full project independently, showcasing critical thinking, clear documentation, and visual communication.