Auto-smooth noisy metrics to reveal trends

Datadog introduces Auto Smoother, an automatic smoothing function that helps users identify trends in their metrics by removing noise while preserving the shape of timeseries data. The algorithm is inspired by Stanford's ASAP (Automatic Smoothing for Attention Prioritization) and uses a moving average to calculate the optimal window size based on two properties: roughness and kurtosis. Auto Smoother offers several advantages over traditional smoothing functions, including automatic adjustment of the smoothing window in real-time and consistent smoothing across multiple timeseries for easy comparison. The feature is now available in Datadog, allowing users to extract valuable insights from noisy data.