如何发现好的研究问题.ppt

What was thought to be complex is really simple Gallego and van Ryzin (1994) show that a complex stochastic dynamic problem (what sequence of prices should be chosen to maximize revenue from a finite set of inventory sold over a fixed period of time) can be well approximated by a deterministic counterpart. What was thought to be simple is really complex The optimal policy to newsvendor problem is known. It is a choice task with random outcomes, and there are a number of established theories for how people make choices under uncertainty 复杂问题简化方法和简单经典问题的新认识 What was expected to be a small effect is really a large effect Jordan and Graves (1995) show that what was assumed to be a small effect (a little bit of flexibility provides only a little benefit) is actually a very large effect (a little bit of flexibility, done right, provides essentially the same benefit as complete flexibility). (A) 2 3 1 2 3 Facility Demand 2 3 1 2 3 Facility Demand (B) 2 3 1 2 3 Facility Demand (C) ? ? 1 1 1 What was thought to be a large effect is really a small effect In the early 1990s there was considerable interest in the use of information technology to improve supply chain performance. The key technologies of the time were bar coding and electronic data interchange Everyone assumed that exploiting the data these systems created and shared across the supply chain would lead to tremendous improvements in inventory performance But Cachon and Fisher (2000) show that what was expected to be a large effect (the reduction in inventory through the use of shared information across the supply chain) was actually a small effect (it is much more effective to move goods faster than it is to be smart with how you move them). What was thought to be a large effect is really much larger Quick response has been shown to provide substantial value to firms (Iyer and Bergen 1997) because of its ability to better match supply with demand. However, Cachon and Swinney (2009) show that ignoring stra