Hypothesis Testing

Hypothesis tests are a ubiquitous part of classical statistics. They often have a very simple objective, such as testing whether two samples of data indicate there is a difference in the means of the underlying populations from which those samples were taken. Despite the simplicity of these aims and questions, hypothesis tests have very practical applications. The question of whether two populations have different means is precisely what we ask when running an A/B test to decide whether the A variant of an e-commerce site has a higher click-through rate, compared to the B variant. As such, hypothesis testing is an important skill to master for any data scientist working with real-world data. Despite the simplicity of the question that a hypothesis test asks, the mathematical machinery needed to run a hypothesis test is full of concepts and nuances that can trip up a new data scientist – concepts such as p-values, degrees of freedom, confidence intervals, Type...