Past Examples of Projects
These projects achieved good to excellent results. The outlines below give some example outputs (but these are only part of the submitted work).
Please don't worry if your idea / project is different to those below - we actually want people to have different ideas rather than all doing very similar projects!
1. Predicting Stock Prices from Core Company Data Using Linear Regression and Machine Learning
In Financial Investment, a key problem for Investors is identifying companies on the stock market that can guarantee a low risk Return on Investment (ROI). Investors select stocks in the capital market based on two main types of analysis, Fundamental and Technical. This team focused on Fundamental analysis which focuses on core factors affecting companies such as earnings, assets, expenses, and liabilities. All these data points are contained within publicly available financial statements.
Data from Companies House API
Modelling with Neural Network
2. Regional Comparison of COVID Impact on Employment
This project investigated how the COVID lockdown impacted on various industries resulting in further widening of the existing disparity between the North and the South of England. The team used statistical analysis methods that capture various significant trend lines. A model was built to subject the data to an hypothesis test. Industries were classified into low contact and high contact, thereby not using a one size fits all approach and reduce bias in our analysis. 21 industries with occupations within them were analysed.
Dataset: Workforce dataset of industries from NOMIS owned by the Office of National Statistics (ONS)
Interactive Dashboards
Visualisations per industry
3. The impact of night-time on female travel habits in the UK
The project objective was to gain a deeper understanding of women’s safety in travel at night-time. The team analysed and illustrated data from the National Travel Survey, seeing how women’s travel patterns differ from men’s and the reasons for this. With this data, they created visualisations telling the story of gender differences in travel.
Tableau Dashboard
JIRA Project Management Board
4. Speech Emotion Recognition System
In this project the team investigated the application of machine learning in the field of speech emotion recognition (SER).
The research identified a shortage of study into the creation of an SER system that can classify emotion from multiple language utterances.
Through a combination of classic and deep learning methods, the team provided a preliminary indication of the feasibility of a multiple language SER model.
Standard SER datasets: EMO-DB, EMOVO, RAVDESS.
Model Overview
Different ML approaches compared
Building on relevant research
5. Household Budgeting
In this project, several suitable datasets were examined and based on some analysis, an application was created to help the most affected households to overcome economic problems and difficulties. According to Clear (2022), there is an urgency for nearly half of adults in the UK to seek for personal finance services. First, sources of family income were identified and then costs that can be reduced were identified based on exploratory analysis and modelling, with the aim to present them in the form of suggestions in a mobile app.
Clear, B. 2022. Half of UK adults need ‘urgent help’ managing their money. Financial Times
Data from Bank of England Household Survey 2011-2021 and ONS expenditure by household composition.
Predictive model for expenditure categories in a household
Mockup of mobile UI
