Unverified50% confidenceSolutionExact time
部分高精度模型会改用负二项分布来修正足球进球数轻微过度离散的偏差
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
5万次模拟2026世界杯：蒙特卡洛模拟与数据科学实战
redditr/learnmachinelearning7/9/2026
Related Claims
UnverifiedDixon-Coles模型在双泊松模型基础上引入了低比分修正项，弥补了0:0和1:0等低分比赛中泊松假设的偏差73% similarUnverified足球的低进球特性使其成为典型的稀疏事件预测问题，一场90分钟比赛平均进球数仅为2-3球67% similarUnverified对于48支球队上百场比赛的赛制，5万次模拟通常可使冠军概率的标准误差控制在约0.2%以内66% similarUnverified假设数据服从高斯分布，最大化似然等价于最小化均方误差66% similarUnverifiedLID切换点准确率通常比非切换点LID准确率低10-20个百分点66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/492131API
curl https://kongchang.com/api/v1/knowledge/claims/492131MCP
get_claim(id=492131)