On Fri, 17 Oct 2003 23:50:29 -0700, "albinali" <[EMAIL PROTECTED]> wrote:
> Hi, > I am using a set of data that I collected using biosensors, the goal is to > analyze the data and to infer the dependencies between the variables to > construct a bayesian network that infers the mood of a person (e.g. happy, > sad...etc). Clearly, the response variable is non-ordinal and categorical. Happy to sad? That looks to me like it *ought* to be one-dimensional, and ordinal. It is not my area, but I am sure that there has been a lot of work in reducing visual displays of emotions to countable dimensions; also, in trying to figure how culturally independent they are. > The model should not include interdependent sensors. I would like to get > some thoughts about using regression for such a task, is it a good choice ? > are there any other possible techniques to use? More importantly, would it > be possible to automate the building of such models? Is anyone aware of any > earlier work in that area?. Thanks. Work on "lie detecting" should also be relevant. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
