|
Abstract:
|
Amazon's obsession with data-driven decision making is very well known and its supply chain is innovating at a tremendous pace. This makes it an exciting time for the supply chain experimentation team as we enable different components of the supply chain to measure the impact of the changes they make. We conduct randomized control trials and on a daily basis we are presented with the question whether a change caused an effect, and what is the nature of this effect. Our automated platform and offline studies produce p-values, point estimates, and confidence intervals. These are the most widely used as well as most misunderstood terms by our customers. In addition to challenges with understanding the nuances of hypothesis testing, we encounter the constant need from our customers to quantify the results with a single estimate and not measures of certainty or the lack of it. This presentation illustrates with examples how our team of statisticians constantly enable our customers to draw inferences from statistical results. We will discuss the challenges faced in doing so and our proposed solutions.
|
