ClementeH/statsbomb-open-data-shots

StatsBomb开放数据——足球射门（xG）数据集是一个包含88,023条射门记录的开放数据集，涵盖了从1958年到2025年共67年的足球比赛数据，涉及21个比赛、48个赛季、308支球队和6,147名球员。数据集特别包含了StatsBomb的预期进球（xG）值作为标签，适用于训练预期进球模型。数据集中详细记录了每脚射门的位置（x和y坐标）、距离球门的距离、射门角度、是否受到防守压力、射门结果（如进球、被扑出等）、射门部位（如右脚、左脚、头球等）、射门技术（如普通射门、凌空抽射等）以及比赛、球队、球员等上下文信息。数据集分为训练集（70,418条）、验证集（8,802条）和测试集（8,803条），且按is_goal进行了分层以保持各分集中的进球率一致。此外，数据集还包含了多个比赛的统计数据，如射门次数、进球数、球员和球队数量等，以及一个简单的梯度提升模型在数据集上的性能基准。

The StatsBomb Open Data — Football Shots (xG) dataset is an open dataset containing 88,023 shot-level records spanning 67 years of football (1958–2025) across 21 competitions, 48 seasons, 308 teams, and 6,147 players. The dataset includes StatsBombs own expected goals (xG) values as labels, making it suitable for training Expected Goals models. Each shot record details the position (x and y coordinates), distance to goal, angle to goal, whether under defensive pressure, shot outcome (e.g., goal, saved), body part used (e.g., right foot, left foot, head), technique (e.g., normal, volley), and contextual information like match, team, and player. The dataset is split into training (70,418 shots), validation (8,802 shots), and test (8,803 shots) sets, stratified by is_goal to preserve goal rates across splits. Additionally, the dataset provides statistics for various competitions, such as the number of shots, goals, players, and teams, along with a benchmark performance of a simple Gradient Boosting model trained on the dataset.