Steps (i), (ii) and (iv) all assume that individual behavior can be related to measurable predictor variables, such as employment status and car ownership. Formal statistical ways of doing this are discussed in Chapters 5 and 14. Of course if you do not own a car, you are unlikely to purchase petroleum from a gas station and so, at an individual level, car ownership is likely to be a good predictor variable. However, ownership data alone provides little indication of frequency of likely visits to the gas station, or the amount spent, both of which are important in identifying the best prospective customers for a location. Other variables such as age and employment status are likely to be helpful in this context, but alone or in combination are rarely likely to provide perfect predictions of behavior. We are unlikely ever to have access to the full range of possible predictor variables, and even those measures that we can ascribe to individuals are likely to be subject to measurement error. We return to the issue of unobserved predictor variables and measurement error in our discussion of uncertainty (Chapters 6 and 15). More fundamentally still, is it acceptable (in predictive terms) to represent the behavior of consumers using socio-economic variables?
- 1 0 年前最佳解答
步(i)、(ii)和(iv)全部假設，單獨行為可以與可測量的預報因子可變物有關，例如就業率和汽車歸屬。 正式統計方式做此在第5章和第14章被談論。 當然，如果您沒擁有一輛汽車，您是不太可能購買石油從加油站然後在一個單獨水平，汽車歸屬可能是好預報因子可變物。然而，單獨歸屬數據提供可能的參觀頻率的少許徵兆給加油站或者花費的數額，其中之二是重要在辨認最佳的潛在客戶為地點。 其他可變物作為年齡和就業率可能是有用的在這上下文的這樣，但單獨或在組合很少可能提供行為的完善的預言。 我們是不太可能得以進入對全方位的可能的預報因子可變物，并且把歸咎我們可以對個體甚而的那些措施可能是受計量誤差支配。 我們在關於不確定性( 第6章和第15章)的我們的討論回到未受注意的預報因子可變物和計量誤差的問題。 更加根本上仍然， (用有預測性的術語 )是否是可接受的代表使用社會經濟的可變物的消費者行為？參考資料： 自己