Should Data Scientists Know The Math Behind Machine Learning Algorithms?
Or just intuition is enough?
It will be useful for data scientists to have a solid understanding of the math behind machine learning algorithms. It allows them to understand what a particular algorithm is actually doing and why they use it!
The maths you actually need for data science include Linear Algebra, Differential Calculus and Probability.
Linear algebra is heavily used in neural networks. Data is represented as matrices and vectors and then we use them with algorithms. In principal component analysis, the algorithm uses singular value decomposition or eigendecomposition which are important topics in linear algebra.
The gradient descent algorithm which is a popular optimization algorithm used to find a local minimum for a function, heavily uses the derivative of the function, which is an important part of calculus. The backpropagation part of the neural network learning process also utilizes the chain rule of calculus.
Uncertainty always exists in machine learning. The algorithms predict the probability of something happening, not the exact thing. Probability is always involved in designing machine learning algorithms. Data scientists need to understand probability concepts to effectively use…
