Philip Wilkinson

Shopping List Application

My lastest project has been the creation of a shopping list application which is currently available on the Google Play Store. The aim of this project was to develop android application development skills. The code repository for this application can currently be found on GitHub and I hope to continue to build on this in the near future. This application currently has:

• The ability to add items
• The ability to delete individual items
• The ability to mark items as in the basket
• The ability to delete grouped items
• A how to use the application page

Medium articles

As part of my interest in Data Science I have regulalry published medium articles on topics and areas of interest. This includes Data Science projects that have used tools and techniques that I have learnt or was interested in. Notable projects include:

• London Conveniece Store Classification using K-means Clustering : Using K-means clustering to be able to group conveniece grocery stores in London according to their immediate surroundings to be able to identify potentially different locational behaviours.
• Random Forest implementation to predict the position of NBA players : Using Random Forest Classification in Python to be able to try to predict an NBA players position based on their stats per minute. The aim was to be able to explore the use of the Random Forest Classifier and to understand whether the NBA has become a "positionless" league.
• A walkthrough of Object Oriented Programming in Python : This articles provides a brief walkthough of Object Oriented Programming practices in Python. The aim of this was to be able to understand Object Oriented Programming by writing a clear and concise guide covering the basics.

Further articles on Medium can be found on my Medium profile.

Head of Science for the UCL Data Science Society (2021-2022)

In July 2021 I took on the position of the Head of Science for the UCL Data Science Society. In this role I built on the work of the previous Head of Science to expand the societys set of workers from 12 to 23 and updated five further workshops. I was supported by 2 Science Executives who helped to create and deliver these workshops throughout the first and second term of the year to the largest student led Technology focused society at UCL. Additionally, all workshops that were delivered during the year were built into a series of Medium articles to increase our exposure. To date, these articles have reached over 200,000 views beyond UCL. A summary of these workshops can be found on Medium here. Workshops that I created include:

• Introduction to Object Oriented Programming
• Introduction to SQL
• Introduction to Advanced Regression Techniques
• Introduction to Decision Tree Classifiers
• Introduction to Random Forest Classifiers
• Introduction to Kmeans clustering
• Introduction to Hierarchical clustering
• Introduction to Dimensionality reduction

And those that I updated throughout the year include:

• Introduction to the fundamentals of Python
• Introduction to sequences in Python
• Introduction to Python logic
• Introduction to Support Vector Machines

Development Planning Unit Alumni Website

Myself and a colleage were commissioned to produce an alumni website for the Development Planning Unit, UCL, to allow their alumni to interact with each other on a dedicated platform. For this I developed a continuous pipeline that updated the website data based on a firebase database of new alumni sign-ups using GitHub action, and created a dashboard that allowed the department to accept or reject new users alongside visualise website usage statistics. This was created over a period of two weeks where the following tasks were completed:

• A python script was created that geocoded the locational information of users into a GeoJSON that could be served up to the website to use in the Mapbox visualisation.
• A GitHub action was set up to load in new data from the firebase database every 12 hours into the Python script, which would then create a refreshed GeoJSON
• A Plotly Dash dashboard was created and hosted on Heroku that would give the department the ability to visualise usage statistics including active users and where they are located. Additional functionality included a downloadable table that contained contact information for users who wished to be included on regular corresponence, and an interactive table that provided the department the ability to accept or reject users based on their profile information which would then feed into the GitHub Actions work in that only verified users would be displayed on the maps created

The website has been successfull running for the past year and successfully hosts over 300 active users.

Hackathons

Since starting my Masters in Smart Cities and Urban Analytics in 2019, and throughout the PhD, I have participated in, and been ranked, in several hackathons. These include:

• Hex Cambridge (2021): My team competed in Hex Cambridge desigining a website to be able to identify and map hostile architecture based on a submission and computer vision tagging algorithm. In this project I was responsible for deployment of the Flask application to the Google Cloud Platform and integration with the AdAstra Database. We were awarded the Wolfram Award and Best use of DataStax Astra. Our submissino can be found on the MLH website here.
• CASA Hackathon (2021): My team competed in the CASA-Hackathon to geocode and analyse residential address data for Dubai for JLL. Our team, on geocoding over 90% of the addresses, was awardded first place. After geocoding the data, we began to explore the data and performed a cluster analysis of sales over the 10 year period of the data. I was responsible for geocoding the data using Open Street Maps and performing the clustering analysis.
• Citadel Data Open (2022) World Finals: Our team competed at the Citidal Data Open world Finals in examining plastic pollution across the world. Out solution utilised a spatial interaction model solution to determine the potential for changes in plastic trade across the world and hence identify future hotspots for increased plastic pollution based on existing plastic pollution policies. Our final report and notebook can be found here. I was responsible for developing and applying the spatial inetraction model to the data to predict plastic trade flows in 2030.