对记录交互数据进行离散化会偏学习决策制定
Discretizing Logged Interaction Data Biases Learning for Decision-Making
AI总结:
本文研究了对非等间隔时间序列数据进行离散化对决策制定模型训练的影响,指出离散化引入了偏差,并提出使用连续时间模型来避免这一问题。
摘要:
时间序列数据通常在非等间隔时间点测量,常通过离散化作为预处理步骤。例如,客户到达时间的数据可能通过将每小时内的到达次数相加来简化,从而生成更易建模的离散时间序列。在本文摘要中,我们展示离散化引入了影响决策制定模型训练的偏差。我们称这种现象为离散化偏差,并表明可以通过使用连续时间模型来避免它。
英文摘要:
Time series data that are not measured at regular intervals are commonly discretized as a preprocessing step. For example, data about customer arrival times might be simplified by summing the number of arrivals within hourly intervals, which produces a discrete-time time series that is easier to model. In this abstract, we show that discretization introduces a bias that affects models trained for decision-making. We refer to this phenomenon as discretization bias, and show that we can avoid it by using continuous-time models instead.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。