Posted by: Margaret Rouse
|“Unfortunately, many default settings on data mining and statistical software automatically create decision trees or rules sets that are preprogrammed to distribute the cases evenly. This can be a huge problem when dealing with infrequent events or otherwise unequal distributions. Another way of stating this is that the program assumes that the prior probabilities or “priors” are 50:50, or some other evenly distributed ratio.” — Colleen McCue|
Today’s WhatIs.com Word of the Day is decision tree, a graph that illustrates a decision-making process by listing every possible outcome as a branch of the tree. (Usually the tree appears to be growing sideways.)
Decision trees are useful for providing team members with a common view of possible outcomes for discussion purposes — but they can also (gasp) be used to assign values to possible outcomes and actually frame what a decision’s consequences would look like.
MBABullshit has a good video illustrating how a decision tree might be used when deciding whether to sell lemonade or candy.