I would suggest focusing on linear algebra first. You might want to consider the Khan Academy Linear Algebra Course to build a reliable foundation before attempting any complex machine learning models.

Like someone mentioned, you definitely need the basics, but unfortunately most beginners ignore multivariable calculus until they get stuck. I had issues with my recommendation logic because I didn't grasp how gradient descent actually optimizes parameters. You'll need to understand partial derivatives at a high level. Use the 3Blue1Brown Essence of Calculus Series since it is free and much clearer than any textbook I've tried.

I totally felt that same deer in the headlights vibe when I first started learning! Honestly, once you get a handle on the linear algebra basics, you absolutely have to dive into statistics and probability. I spent months trying to build a recommendation engine and it was a total disaster until I finally understood how likelihoods work. Understanding the probability of someone liking a specific genre based on their history is pure stats! I love how it turns those scary symbols into actual logic. It makes everything feel so much more manageable. You also need a solid grasp of basic calculus, specifically derivatives. Dont let the name scare you off tho!! Its basically just finding the slope of a line to see if your model is getting more accurate. I remember the exact moment it clicked for me while I was using Google Colab Free Tier to run my first training loops. Watching the loss values drop because of gradient descent math was amazing! If you want a more structured path, I highly recommend looking at the DeepLearning.AI Mathematics for Machine Learning Specialization via Coursera. You can audit the videos for free, and it methodically explains the why behind every concept. Focus on the stats first, then move to derivatives. You are gonna do great with that bookstore app, it sounds like a fantastic project!!