[R] MNP,multinomial probit model

Ozlem Ipekci ozlemipekci at gmail.com
Thu Apr 19 17:05:29 CEST 2007


I am an undergraduate student in Isik University, in Istanbul/TURKEY.  For
my term project, I am studying on the relationship between the prices and
market shares of 3 automobile models.  I need to find estimates for the
parameters so that I can continue my study.  I found about your MNP package
application on detergent and Japan-election data.


I have found about R, just a week ago.  I am totally a stranger to the
package, but I am desperate to learn how to implement it.
I have installed MNP, after installing the R, but I could not form the
correct prompt line to fit my data to Multinomial Probit model (MNP).  My
problem is as follows:
I am studying on the relationship between the prices and market shares of 3
automobile models.  I want to find estimates for the model parameters, which
MNP finds.
I consider three primary automobiles corresponding to the same segment;
FIAT-Grande Punto, CITROEN-Clio 3 and VW-Polo.  The data I have is 12-months
prices for each automobile and their market shares at these months
(PricePunto, PriceClio, PricePolo, MarketSharePunto, MarketShareClio,
MarketSharePolo).
By looking at the examples on detergent-choices and Japanese election
applications available with the MNP package, I figured something like;

> resX <- mnp(cbind(MarketSharePunto, MarketShareClio, MarketSharePolo) ~ 1,
choiceX= list(PPunto=PricePunto, PClio=PriceClio, PPolo=PricePolo), CXnames
= "price", data = rowdata, verbose = TRUE)
or
> resX <- mnp(cbind(MarketSharePunto, MarketShareClio, MarketSharePolo) ~
PricePunto + PriceClio + PricePolo, data=rowdata, verbose=TRUE)

But these two lines do not fit the model to my data; gives errors that I
even cannot understand.
I could not find the details on MNP function.
If I can work this out, I will apply it to another dataset, which involves
different automobiles for a longer period of time.
If anyone can help me on this, I would be very pleased.
Thank you!

Ozlem Ipekci
(I attached the dataset incase)
-------------- next part --------------
PricePunto PriceClio PricePolo MarketSharePunto MarketShareClio MarketSharePolo 23250.0	23900.0	23700 0.007665314 0.009979841 0.009730967
23477.0 24590.0 23250.0 0.007336273 0.006573795 0.006862301
25220.0 25410.0 35350.0 0.00478962 0.007396629 0.001667273
26330.0 26440.0 25360.0 0.003884524 0.005164651 0.00304582
26980.0 27070.0 26240.0 0.004984724 0.003617945 0.005828911
26980.0	25070.0	26490.0	0.004000873 0.007128828 0.004510075
26980.0 25600.0 26490.0 0.006639356 0.006215567 0.003531572
24480.0 25600.0 26490.0	0.008695082 0.005873281	0.002690554
24480.0 25600.0 26640.0 0.004983711 0.005372052 0.005307329
25750 26100 25850 0.002806006 0.004626119 0.005232823
25760.0 26100.0 25850.0 0.004473623 0.003892633 0.003311643
24760.0 24850.0 25850.0 0.007923077 0.006730769 0.003692308


More information about the R-help mailing list