My MSc dissertation: designing and validating a quantitative metric that scores how good a university timetable is turning fuzzy student preferences into a measurable number via a structured survey and data analysis. Supervised by Dr Anthony Kleerekoper, MMU.

Compared a baseline CNN against ResNet18 and ViT-B/16 in PyTorch on a ~20k-image dataset, then shipped the best model as a live Gradio app.

Classifying tumours as benign or malignant from diagnostic features, comparing six algorithms with cross-validation. SVM came out on top; feature scaling gave a clear lift.

A 10-stage regression pipeline on ~400k UK used-car listings encoder benchmarking, RFE feature selection, tree ensembles and a final voting ensemble, with SHAP, PCA and Isomap.

Benchmarked five regression algorithms linear, polynomial, decision tree, random forest and SVR to predict a combined-cycle plant's net hourly energy output from environmental readings.

A K-Nearest Neighbours classifier predicting whether income exceeds 50K from demographic and work features, with feature scaling and optimal-K tuning.

A Kafka + Spark Structured Streaming pipeline processing Amazon Electronics reviews in real time, plus an OpenMP/MPI parallelisation study benchmarking scaling across cores.

An interactive Power BI dashboard turning raw sales data into decisions revenue by product, category and region, monthly trends and KPI monitoring, built on a clean data model with DAX.

An interactive Tableau dashboard exploring sales performance across categories and regions calculated fields, filters and parameters that surface which segments outperform and when.

An interactive Excel dashboard on Blinkit grocery sales Power Query, pivot tables, slicers and VBA revealing top outlets and categories. Tier-3 outlets and low-fat products drive the most revenue.