[[Notes]] - Topics: [[Learning]] - People: [[Thu Vu]] - Source: https://www.youtube.com/watch?v=A8Abf3u0ZIs --- ## Summary - "[[What you need to learn really depends on what you want to do]]". - [[You don't need to be an expert to be effective]]. - [[Simplicity is abstractions done right. Complexity is abstraction, unmanaged.]] - [[Manage messiness by first managing yourself]]. - [[Ultimately, problem-solving is about not giving up]]. ## Notes - The 3 common areas for data science include, linear algebra, calculus and statistics and probability. - Statistics is more science than math because it requires real world data. - The difficulty in learning something can often be measured by how abstract something is. - Simplify concepts by reducing abstractions. Intentional abstractions can also help with understanding. - [[Simplicity is abstractions done right. Complexity is abstraction, unmanaged.]] - It's more useful to have a broad and solid understanding of your area of study and work. - [[It's like driving a car. What's important is your ability to get from A to B safely, not your encyclopedic knowledge of how cars are put together.]] - [[You don't need to be an expert to be effective]]. ### Linear Algebra - Linear algebra encompasses vectors and matrixes and the operations performed on them. - Principal component analysis uses singular value decomposition to present the data in fewer dimensions. - Linear Algebra is the backbone of neural network calculations. - Essential linear algebra concepts - dot product - Matrix multiplication - Matrix factorization / LU decomposition - Eigen vectors, Eigen values - Single value decomposition ### Calculus - Calculus is the mathematical study of continuous change - Essential calculus concepts: - Limits - Derivative of a function - Integrals - Partial derivations - Chain rule Calculus is seen in performing back propagation algorithms for neural networks ### Statistics and probability - Machine learning is like fancy statistics. - Essential statistics and probability concepts: - Mean mod, quantiles, standard deviation variance - Covariance, correlation - Conditional probability - Probability distributions - Sampling - Hypothesis testing - "[[What you need to learn really depends on what you want to do]]". - Action and understanding are the goals. Learning is how you achieve them. ### Discrete math - Discrete math is used a lot in computer science. - Examples include sets, counting functions, basic data structures and big O notation. - Always remember your goal and focus on what is essential. - [[Explore tangents but never get lost in them]]. - Get the fundamentals right. - [[Manage messiness by first managing yourself]]. - Don't be overwhelmed by the dread of mess. Be grateful to be exposed to something new. - [[Ultimately, problem-solving is about not giving up]]. - Divide and conquer your learning. Be creative with it. - [[Breaking things down surfaces what you know and what you don't know.]] - The giant staircase of hard is just 100 steps of easy. ## Original ![[How to learn math for data science (and stay sane)-1.jpg]] ![[How to learn math for data science (and stay sane)-2.jpg]] ![[How to learn math for data science (and stay sane)-3.jpg]]