Blog roadmap
Introduction
Welcome back! I haven’t had the time to write much these past few months. I started a new job as a junior data scientist for IKEA, one of my favorite companies (who here hasn’t spent a good afternoon browsing an IKEA store?). It has kept me quite busy, especially since the first few months have consisted of an Accelerator Programme filled with various trainings in all things DS, ML, MLOps, and anything needed to do the job.
After a particularly reinvigorating Christmas break, I once again felt the urge to write about all the things I have been learning and working on. So here is my roadmap for the coming days, weeks, and overall year, of what I’d like to write about.
Roadmap
Timeseries forecasting
During my accelerator training, I was lucky to get a thorough introduction to timeseries: how to work with them, how they differ from your typical machine learning problem (and here just like in physics, causality always comes to bite you) and what tools to use. What I particularly enjoyed was the focus of the excellent library Skforecast, co-maintained by one of our own IKEA coworkers (hey Javier 👋). I had previously used this library for one of my projects and I had really enjoyed how easy it was to get stuff done with it. I also whole-heartedly recommend their documentation: it’s a fantastic place to learn how to properly handle timeseries data with numerous examples (for instance, it contains the best explanation of backtesting techniques I have managed to find).
Fresh from this week of learning, I’d like to take another look at my energy forecasting project, maybe update it a bit and write about it here.
Kaggle competitions
When I started learning about data science, I found it illuminating to try my hand at Kaggle competitions, both as a way to practice what I read about, but also because there are so many well-explained notebooks which can serve as further learning material. It has almost been a year now since the start of this journey. I have spent a lot of time learning all sorts of best practices as well as simple (and not so simple) algorithms, and I have been itching to test all of this knowledge on some Kaggle competitions. It’s a nice way to see where I’m at. And lucky me, I still have a week of holiday to occupy myself with. So stay tuned for that!
New personal projects
As anyone would tell you, there is no better way to learn than to practice. After working all day, it’s not easy to open my computer again in the evening to work on side projects. But there are a couple of ideas that have been stewing in my head for a bit now and I’m eager to try them during this break. The Accelerator Programme will be focusing on computer vision and GenAI soon, so it will be a nice time to warm up my NLP and deep learning muscles!
Among these ideas, I’ve been thinking of creating a simple pipeline for calorie counting! It’s a simple habit which I have found to be a great way to stay healthy, but it’s not always easy to account for everything. Usually, I just log every item in the FOSS Android app Energize. However, for complicated recipes, it’s difficult to log every single item from just the description. Especially once you want to repeat meals. Instead, I was toying with the idea of making my own webapp, with similar features to the Android app but with the added bonus of using a model to parse recipes or ingredients list.
Summary
I’m not one to make New Year resolutions, mostly because I never stick with them, but I’m quite excited this year to work more on side projects and as a byproduct, to write about them! I have plenty to stay busy, and I have some travel goals too which will take priority, but hopefully I can finish all these projects before I write another New Year article.