Using apartment data scraped from ingatlan.com to predict real-estate prices in Budapest using beautifulsoup, and polynomial regularized regression models.

Using music streaming data from KKBOX and metadata from Deezer to predict user behaviour, leveraging state-of-the-art classification techniques, e.g. logistic regression, random forest and xgboost

Using Twitter comments of Fortune 500 companies for finding business-relevant clusters using web-scraping, natural-language-processing, NMF modelling and hierarchical clustering.

Using Tableau visualisations and Convolutional Neural Networks to predict future cell-tower traffic patterns.