[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: predictive regression with mutiple dependent variable

From   "Martin Weiss" <>
To   <>
Subject   st: RE: predictive regression with mutiple dependent variable
Date   Wed, 24 Sep 2008 13:17:37 +0200

See -h mlogit-. BTW, why would you want to reduce the information of a
continuous variable in 1 and 2? In 3., it is clear that industries should be
represented by a discrete measures, I 1. and 2. It is more difficult to see
why you want the quintiles and not the variable itself...


-----Original Message-----
[] On Behalf Of Richard Harvey
Sent: Wednesday, September 24, 2008 12:37 PM
Subject: st: predictive regression with mutiple dependent variable

Hi all,

I am trying to build a model for predicting the probability of a firm
falling into either one of four seperate classes (the dependent
variable) using several independent variables.

Lets say the groups are G B U and S.

now the independet variables consists of

1. a variable which can take a value of 1 to 5.  This variable is
created by sorting all the units based on this variable and putting
then into quintiles.So this number represents the quintile it belongs

2. is similar to variable 1, can take of 1 to 5, this is created on
quintiles of another measure

3. represents the industry (there are 5 broad industry groupings)

4. 3 other continous variables

5. Some of the variables in 3 may have been affected by the state of
the economy during ceratin periods. So time or state of economy should
somehow be factored in. But  i am not sure how a  varaible for year
will work in the context of a prediction model. I suppose it will then
have to be a variable representing the state of the economy.

and the  rest are all continous variables

What model should i be looking at?

thanks for your time
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index