Datadog/BOOM

BOOM（可观测性指标基准）是一个用于评估复杂可观测环境中模型预测任务的大规模真实世界时间序列数据集。它由Datadog平台收集的实际指标数据组成，涵盖了基础设施、网络、数据库、安全和应用程序级别的指标。BOOM数据集具有高度的不规则性、结构性复杂性和重尾统计特征，包括零通货膨胀、高度动态模式、复杂的季节性结构、趋势和突变、随机性、重尾和偏斜分布以及高基数等特点。

BOOM (Benchmark of Observability Metrics) is a large-scale, real-world time series dataset designed for evaluating models on forecasting tasks in complex observability environments. It consists of actual metric data collected from Datadog, covering metrics from infrastructure, networking, databases, security, and application levels. The dataset exhibits features such as zero-inflation, highly dynamic patterns, complex seasonal structures, trends and abrupt shifts, stochasticity, heavy-tailed and skewed distributions, and high cardinality.

搜集汇总

背景与挑战

背景概述

BOOM是一个大规模的真实世界时间序列数据集，用于评估复杂可观测性环境中的预测任务。数据集包含约3.5亿个时间序列点，涵盖多个领域的指标，具有不规则性和结构复杂性等特点。