停车场剩余车位短时预测误差分析数据集

本数据集聚焦停车场剩余车位短时预测的误差分析，覆盖路内停车、地面/地下停车场、立体车库、P+R换乘停车场及商业综合体配建停车场等类型。数据内容包括预测任务元数据、输入特征、模型预测值、实际观测值、误差度量（绝对误差、相对误差、偏差方向）及误差根因标签等。适用于预测算法评测、失效根因分析、动态定价优化、停车诱导系统改进等场景。

This dataset is dedicated to error analysis for short-term prediction of available parking spaces, covering various parking types including on-street parking, ground and underground parking lots, mechanical parking garages, park-and-ride (P+R) lots, and parking lots attached to commercial complexes. It contains prediction task metadata, input features, model-predicted values, actual observed values, error metrics (absolute error, relative error and deviation direction), as well as error root cause labels. It is applicable to scenarios such as evaluation of parking prediction algorithms, root cause analysis of prediction failures, dynamic pricing optimization and improvement of parking guidance systems.

搜集汇总

背景与挑战

背景概述

该数据集专注于停车场剩余车位短时预测的误差分析，覆盖了路内停车、地面/地下停车场、立体车库等多种停车场类型。数据内容包括预测任务元数据、模型预测值与实际观测值的误差度量及根因标签，适用于预测算法性能评估、失效原因分析和停车诱导系统改进等应用场景。