Isaac Aderogba

Changelog #1 | Compounding yourself

November 22nd, 2020

I'm rounding up my time at ToDesktop to speed up my transition to AI. While not the best move financially, I fell off the exponential curve of technical improvement.

In keeping with the startup ethos, I'll be posting a weekly changelog and shipping projects frequently. I've got ~5 months of runway, let's start building 🛠.

Training a NER model to identify food entities

In this project, I trained spaCy to identify FOOD entities from sentences like:

  • I decided to have chocolate ice cream as a little treat for myself.

  • I had a hamburger and chips for lunch today.

  • I ordered basmati rice, leaf spinach and cheese from Tesco yesterday.

This is part of a larger side project to make nutrition tracking as effortless as having a conversation.

Visualising earnings by college degree

In this project, I analysed job outcomes from different college degrees. I answered questions like:

  • How are median earnings impacted by degree popularity?

  • How are unemployment rates impacted by degree popularity?

  • How are median earnings impacted by degree gender balance?

This project gave me some practice with matplotlib. It's part of a wider course that's giving structure to my self study.

Creating a value metric for EBay used-cars

In this project, I created a value-to-price metric to rank different used-car models.

This project was mainly focused on data cleaning. It gave me some practice with pandas and numPy.

Investigating Hacker News post engagement

In this project, I analysed Hacker News data to understand what posts drive more engagement.

This project restricted me to using Python and its core libraries for data analysis. This is the project that introduced me to this style guide.

Examining profitable apps for App Store and Google Play

In this project, I explored what types of apps attract more users on the App Store and Google Play.

This was my very first project! Just to get me up to scratch with Python programming and structuring data science projects.