什么是预测

  • Forecasting is the process of making predictions of the future based on past and present data along with analyzing the trends. 预测，基于历史和当前数据形成对趋势的预估，进而预估未来情况的技术。
  • forecasting analytics is the extensive use of data and quantitative models as well as evidence-based management and judgment so as to produce alternative point and density estimates, paths, predictions, and scenarios for the future.
  • 维基百科2018年关于 forecasting 的描述：
    • Forecasting is the process of estimation in unknown situations. Prediction is a similar, but more general term, and usually refers to estimation of time series, cross-sectional or longitudinal data. In more recent years, Forecasting has evolved into the practice of Demand Planning in everyday business forecasting for manufacturing companies. The discipline of demand planning, also sometimes referred to as supply chain forecasting, embraces both statistical forecasting and consensus process.
    • 理解 1：estimating in unknown situations 所以你武器只有过去，以及未来能够和过去有多么相似。即所有预测技术的统一前提：未来和历史相似。
    • 理解 2：这句话区分了 forecast 的 predict，但是似乎仍然没有太大的差异，所以目前我个人还是混用这两个词汇。可能 predict 更定量，forecast 定量定性都包含？
    • 理解 3：最主要的应用场景就是供应链需求预测。

预测和因果性

  • 在预测的场景里，我们不关注 因果性 这个圣杯。

  • 只要能预测准确，就可以，不需要关注因果性。

    预测和统计学的区别

  • 统计学家认为数据之下暗藏一个模型，它们的任务是将这个模型应用到未来。

    • In essence, in statistics we try to find the model that best fits the data. And since we expect history to repeat itself, we project it and we are happy.
  • 预测专家则关注哪一个模型预测得最好。留下 20% 的数据或者与预测时段等长的数据，用于回测。所以预测专家给出的结果不一定是统计学家给出的最 fit 的结果。

不确定性永远伴随着预测

  • 告诫：“ . . . make sure you do your best to get an as-accurate-forecast-as-possible, 3 and learn to live with the uncertainty that will inevitably come with this forecast . . . ”
  • There is no guarantee that the model that forecasts best, will keep on forecasting best. 预测模型不能保证自己永远预测得最好。
  • However, it still produces on average better forecasts than the model that fits best! 然而，预测模型总是可以给出平均水平最优的预测。

预测的局限性

  • Evaluating different forecasts based on historical accuracy is helpful only if historical patterns continue into the future. 只有历史模式能够延续到未来时，基于历史准确率挑选出来的预测方法才有作用。
  • Good judgment and knowledge about business conditions that might affect the value of the variable to be forecast also have to be considered carefully when selecting a method.

预测的产出

  • point-forecast
  • forecast/prediction intervals

预测技术类型