Research and Projects
Healthcare Research: Alzheimer’s Disease (BIOCARD Project)
Tools: Python, R, Statistical Modelling
Analysed longitudinal MRI and biomarker data to identify patterns of neurodegeneration and disease progression.
-
Applied mixed-effects models to study amyloid and tau accumulation
-
Investigated genetic and medical risk factors (e.g., APOE4, medications)
-
Mapped progression trends across brain regions
Outcome: Developed insights into disease progression patterns, contributing to ongoing research in neurodegeneration.
Early Retirement Prediction Model - ELSA Wave 7
Tools: Python, XGBoost, CatBoost
Built machine learning models to predict early retirement using demographic, financial, and behavioural data.
-
Achieved ROC-AUC of 0.67
-
Incorporated quality-of-life indicators (CASP framework)
-
Published findings as a research preprint
Outcome: Provided insights into key drivers of early retirement decisions.
Post-Retirement Well-being Analysis
Tools: Python, Statistical Analysis, Clustering
Evaluated how retirement timing influences quality of life using the English Longitudinal Study of Ageing (ELSA).
-
Analysed cross-sectional data of adults aged 50+ using the CASP-19 well-being framework
-
Applied non-parametric statistical tests to compare early vs. statutory-age retirees
-
Used unsupervised clustering to identify distinct post-retirement profiles
-
Explored the role of financial, social, and health factors in shaping well-being
Outcome: Found that early retirees tend to report a higher quality of life, particularly in autonomy and pleasure. Identified key drivers such as relational closeness and financial preparedness.
Premier League Prediction Model
Tools: Python, Machine Learning
Developed predictive models using team performance and financial data to forecast league outcomes. The project involved applying feature engineering and statistical modelling techniques to capture underlying performance trends and key indicators across teams.
Outcome: Demonstrated strong predictive modelling capabilities within a sports analytics context.
Sentiment Analysis using BERT- Vision Pro v/s Quest 3
Tools: Python, BERT (NLP), Looker Studio
This project focused on analysing over 12,000 YouTube comments to compare audience reactions and engagement for two competing products. Using BERT-based sentiment analysis, I evaluated large-scale textual data and translated the findings into an interactive dashboard, making the insights easily accessible to a broader audience.
Outcome: Identified clear differences in audience perception and engagement using large-scale natural language processing techniques.
Analyzing Gender Bias in TED Talk Audience Sentiment Using NLP
Tools: Python, NLP, Looker Studio
In this project, I analysed over 2,600 YouTube comments across 12 TED Talks to investigate differences in audience sentiment toward male and female speakers. By applying natural language processing techniques, I quantified sentiment patterns and compared reactions across similar topics to uncover underlying biases.
Outcome: Revealed measurable disparities in audience sentiment, highlighting potential gender-based differences in audience perception.