Statistics vs. Machine Learning

Statistics and Machine Learning are two different, but related fields. Both focus on data and both use data to estimate parameter values, but they differ in purpose and methods.

Statistics is about understanding data and testing hypotheses. It attempts to draw conclusions from the data. For example: "Does the minimum wage affect the employment rate? If so, what is the direction and size of the effect?"

Machine Learning focuses on larger, more complex data problems It makes predictions, with an eye towards automation. For example: "Given a sentence, predict its translation." or: "Given an image, describe its content."

To estimate parameter values, Statistics uses calculus. There's a mathematical reason why, when estimating the average, we focus on the sample mean (the sum of the observations divided by the number of observations).

By contrast, Machine Learning uses gradient descent, which is essentially a "trial and error" method. Starting with randomly initialized parameter values, it compares a batch of predictions to the true value, then it adjusts the parameter values in the direction that most reduces the error.