Projects
Selected data science work, written as case studies with context, methods, and outcomes.
-
SMS Spam Detection Using NLP
Nov – Dec 2023 · Natural language processing and cloud deployment A machine learning project that classifies SMS messages as spam or legitimate text, then packages the trained model for serverless inference on AWS. Outcome An SGD classifier reached 97.9 % accuracy and a spam F1 of 0.919 on 1,115 held-out messages, then shipped as a Docker image behind AWS Lambda and API Gateway. Pythonscikit-learnNLTKDocker +3 more Read case study -
Predicting Student Success
Oct – Nov 2023 · Education analytics and multiclass classification A student outcome prediction project using higher-education enrolment, demographic, socioeconomic, and academic performance data. Outcome Random Forest scored best of ten classifiers on the validation set, with 76 % accuracy and a macro-averaged F1 of 0.68 across three student outcomes. Pythonscikit-learnFlaskwaitress +2 more Read case study -
Mathematically Modelling Microbial Interactions
Sept 2022 – May 2023 · Mathematical biology, Bayesian inference, and dynamical systems A final-year project modelling interactions between Staphylococcus aureus and Pseudomonas aeruginosa in co-culture. Outcome Bayesian fits of a two-species Lotka-Volterra model, plus a bifurcation analysis, showed the switch from co-existence to P. aeruginosa out-competing S. aureus. RStanHamiltonian Monte CarloMCMC +2 more Read case study -
Modelling Epidemic Spread in a Dynamical Network
Jan – Apr 2022 · Epidemic simulation and dynamic network modelling A third-year project studying how changing network structure affects SIR and SIRS epidemic simulations. Outcome Across 5,450 simulated outbreaks, epidemic risk rose with the number of partners per person and fell with longer partnerships. REpiModelSIR/SIRSdynamic networks +1 more Read case study