COMP20008 - Elements of Data Processing
Data processing and analysis in Python on Victorian road crash data, from cleaning and text analysis to clustering and predicting injury outcomes.
Both assignments in this subject used Victorian road crash data (people, vehicles and accidents), working in Python with pandas, matplotlib, NLTK and scikit-learn.
The first was about cleaning, exploring and visualising the data: grouping people into age brackets and comparing seatbelt use across them, and processing the free-text accident descriptions (removing stop words, word frequencies and word clouds) alongside when in the day accidents happen.
The second asked a research question: how do accident type, seatbelt use, seating position and vehicle type relate to injury outcomes? It was answered with correlation analysis, including normalised mutual information, k-means clustering with the elbow method to choose k, and decision tree and k-nearest neighbours classifiers to predict injury outcomes, evaluated with confusion matrices.