Workforce population modelling: Population by convolution
本文提出用卷积方法预测劳动力人口变化，将人员流入与生存时间分布卷积得到未来流出，进而计算人口水平，该方法比传统马尔可夫方法更通用且计算快速。
In this paper we pursue the premise that personnel inflow into a workforce population maps to the future outflow via the survival time distribution. This directly leads to a population by convolution approach, where the survival time distribution is convolved with the intake profile over time to find the future outflow; the population level is then the cumulative sum of the difference between the intake and outflow. This approach is simple, computationally fast, and provides the expected (mean) population level. We contrast convolution with the standard state-of-the-art Markovian approach, which assumes a memoryless survival time distribution, noting that determining population by convolution is a generalization.