tbh if you are looking at warehouse data and shipping delays, you really gotta prioritize linear algebra first. it is the actual engine that makes the code work, whereas stats is more about the logic. i have been super satisfied with how much easier things get once you view every dataset as a multi-dimensional matrix. if you dont understand dimensions and dot products, you are gonna get stuck with shape errors in every single model you try to build. it felt great when i finally figured out how tensors actually flow through a system. for your 5-hour window, Wellesley-Cambridge Gilbert Strang Linear Algebra and Learning from Data 1st Edition is basically the gold standard. it focuses on the stuff that actually matters for ML rather than just abstract proofs. it works well for self-study because he explains the mechanics behind the numbers. once you get the hang of basic matrix operations, you will see your shipping logs way differently.

Stumbled upon this today and figured id weigh in as someone who prefers the safe route. Given your limited time at the warehouse, focusing on probability and statistics is definitely the most reliable way to avoid making expensive mistakes with shipping predictions. I have been very satisfied with how this approach minimizes risk. If you compare it to jumping into calculus, the stats route is much more practical for someone on a shift schedule. Ive found that O'Reilly Practical Statistics for Data Scientists 2nd Edition is a solid choice because it focuses on what actually works well in a business setting without all the academic fluff. It is very efficient for those with tight schedules. On the other hand, something like Springer An Introduction to Statistical Learning with Applications in Python is great if you want a more methodical, rigorous foundation. I have no complaints with the Springer text, though it might take you longer to get through. Both offer different levels of safety. Springer is more thorough but O'Reilly is faster for immediate warehouse applications. Calculus is important eventually, but for warehouse data, you mostly need to understand distributions and variance so you dont end up with a model that looks good on paper but fails in the real world. Cost considerations are real when youre talking about logistics. Tbh, being happy with your results usually comes from knowing your data is solid, not from manual derivatives. Stick to the applied stats first. It is a much safer bet for your 5-hour window. No need to stress over the heavy math yet.

Re: "Stumbled upon this today and figured id weigh..." - while the others make good points, be careful about sinking all your time into academic theory when you only have 5 hours a week. If you spend months on pure linear algebra, you might get burnt out before you even touch a shipping model. Honestly, for warehouse work, you need to prioritize reliability over complex math. I'd suggest a slightly different path to save some cash and time:

• Dont buy those expensive $100 textbooks yet. Grab something like Dover Publications Introduction to Probability since its way cheaper and covers the essentials for understanding distributions.
• Focus on Descriptive Statistics first. If you dont understand your warehouse lead times and variance, no amount of calculus will fix a broken model.
• Check out No Starch Press Statistics Done Wrong to see how people mess up real-world data. Its a great way to learn what to avoid without spending a fortune on a degree. I would suggest staying away from the heavy proof-based stuff for now. You want your predictions to be reliable, not mathematically perfect but useless in practice. Just keep an eye on your budget and dont buy into the hype that you need a PhD-level math stack to start being useful at work.

Came here to say the same thing lol. Great minds think alike I guess.

ngl, it's kinda disappointing how much hype there is around needing crazy calculus right away. i wasted so much time on derivatives when i first started and it really wasnt as good as expected for actual practical work. i had issues with my early models just giving weird, unrealistic predictions because i didnt understand the underlying data distribution, which was a huge bummer. if i were you, i would focus 100 percent on statistics and probability first. linear algebra is cool for the heavy mathy side of deep learning, but for warehouse logistics and shipping delays? stats is your best friend. you need to know if a delay is a random outlier or a real trend that needs fixing.

• probability distributions: you will use these to model wait times constantly
• hypothesis testing: helps you see if your new model actually makes better guesses
• regression analysis: basically the bread and butter of predicting shipping dates calculus is technically the foundation, but honestly you wont be using it for your daily tasks. id stick to a solid textbook like Pearson Statistics for Business and Economics 13th Edition instead of drowning in vector spaces right now. it sucks to realize you need more math, but stats is the safest bet for reliable results in your specific job. just dont rush it... you dont want your warehouse manager mad cuz a package was a week late lol.