How AI is Changing Data Collection

Many internet companies are changing how they collect data these days. They are following the principle of “data minimization,” which means they’re collecting less data. This can make A/B testing less effective though. A popular technique for experiments with multiple observations is to aggregate data for each unit. But to get exact quantile estimation, you need all the observation-level data. In this article, we will discuss a new method for analyzing Quantile Treatment Effects (QTE) using histogram aggregation. We will also talk about how we can ensure formal privacy guarantees by using differential privacy.