Projects & Blogs

An exploration into traditional investment strategies, comparing the performance of 'Buy and Hold' and 'Dollar Cost Averaging' across volatile stock datasets.

An in depth exploration of stock market data through feature engineering and pattern recognition. This study analyzes historical price changes and volatility to identify trends across diverse companies using custom statistical metrics and datetime features.

We compare the Romanian dataset with our own surveyed dataset and explore the relationship between the sociodemographic and financial behaviour, with financial well being and financial literacy.

Part 2 of the analysis focuses on High Risk Financial Vehicles (HRFV), data collected with the Malaysian dataset were inspired by and referenced from a Ghanaian study. We explore how financial literacy influence perception and participation in activities like sports betting and cryptocurrency.

This project conducts a detailed evaluation of popular machine learning algorithms and their performance characteristics. It benchmarks Naive Bayes, Random Forest, and k Nearest Neighbors across multiple datasets ranging from simple iris data to complex geospatial and text categories. The analysis explores the relationship between hyperparameter tuning and model efficiency while providing quantitative results on accuracy and execution time.

I created a machine learning model to categorize types of Italian wine. This project features a custom implementation of the k Nearest Neighbour algorithm. I also developed the nested cross validation logic without using external libraries. The work explores model stability when handling data with Gaussian noise.