Homepage

Is there such a thing as 'Oscar-bait'?

A look into English-speaking movies, award winner and nominees from 1934 to 2017 using data from IMDb, Kaggle and self-curation.

An exercise with a very fruitful and teaching preprocessing part. The multi-million row IMDb datasets, filtered and enhanced with webcrawling will probably inspire more analysis later on.

Read

Link (R Markdown)

Analysis on Taylor Swift - reputation

Using audio data from Spotify and lyrical sentiment analysis using tidyText to compare Taylor Swift's latest album to her previous work

First exercise without a ready-to-go dataset. First time working with APIs and pipes (%>%). Further practice with ggplot2.

Link (R Markdown)

Exploring Retail Sales

Exploratory data analysis on sales of Rossman stores in Germany

A learning practice in visualization using ggplot2, and exploratory data analysis.

Link (R Markdown)

House Pricing Prediction

Predicting sale prices for houses, given labeled data for previousy sold homes.

A regression problem solved using XGBoost & 10-fold cross validation.

Link (R Markdown)

McKinsey Hackathon - Stroke Prediction

Using labeled patient data to predict possibility of a stroke for individual patients.

My submission to McKinsey's 24-hour online hackathon. A probabilistic classification problem solved using conditional random forest. Earned an AUCROC score of 0.847, while the winning submission got 0.860.

Link (R Markdown)

Titanic Survivor Prediction

Using labeled passenger data to predict who survives the disaster

A "Hello World" project to data science. A classification problem with three seperate solutions using logistic regression, decision tree and random forest.

Link (R Markdown)