2:59 p.m.
Hi,
Suppose I have a data set of 200 cars where the observation labels are
the make (Chevy, Honda, BMW, etc.). Say there are 10 makes. I want to
create a variable, say ID, that identifies the make so I don't have to
rely on the observation label. The purpose is to later create dummies
for the makes to include in a regression model and create boxplots (one
for each make's price, all shown in one plot window) of price. How can
I do this? Or is it better to just use the observation labels and
laboriously write if statements?
Thanks,
Walt
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
_____________________________________________________
--
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
11:17 p.m.
Don't know how others view it but seems to me that 200 obs
for what you are seeking to do doesn't make for good regression
inferences....but your call.....best wishes...
Dr RJF Hudson Qld Australia
rjfhud(a)powerup.com.au
----- Original Message -----
From: Data Analytics Corp.
To: gretl-users(a)lists.wfu.edu
Sent: Saturday, May 15, 2010 12:59 AM
Subject: [Gretl-users] creating ID variables
Hi,
Suppose I have a data set of 200 cars where the observation labels are
the make (Chevy, Honda, BMW, etc.). Say there are 10 makes. I want to
create a variable, say ID, that identifies the make so I don't have to
rely on the observation label. The purpose is to later create dummies
for the makes to include in a regression model and create boxplots (one
for each make's price, all shown in one plot window) of price. How can
I do this? Or is it better to just use the observation labels and
laboriously write if statements?
Thanks,
Walt
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
_____________________________________________________
--
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
8:54 a.m.
Data Analytics Corp. schrieb:
Hi,
Suppose I have a data set of 200 cars where the observation labels are
the make (Chevy, Honda, BMW, etc.). Say there are 10 makes. I want to
create a variable, say ID, that identifies the make so I don't have to
rely on the observation label. The purpose is to later create dummies
for the makes to include in a regression model and create boxplots (one
for each make's price, all shown in one plot window) of price. How can
I do this? Or is it better to just use the observation labels and
laboriously write if statements?
AFAIK the term "observation labels" in gretl terminology is used to
describe individual datapoints, not as a field to hold certain defining
characteristics which is effectively a variable. So, starting from a
spreadsheet file, if you put the "make" variable in its own column (with
character string values) then gretl should automatically code it with
numerical values and notify you when you import that file (in .csv
form). Then it seems to me that you will have what you need: You can
restrict the sample to those obs where the ID equals a certain number
code. If you want you can also create dummies like so:
series hondadummy = (ID=5) # example with code 5 for Honda
This can also be done for all makes in one step in the GUI I think: try
Add -> create dummies for selected discrete variables.
HTH,
sven
11:27 a.m.
Hi Sven,
I did have make in its own column, the first. I'll try to put it in
another column position and see what happens. If gretl assigns numerics
to the makes, how do I know what numeric goes with what make?
Thanks,
Walt
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
_____________________________________________________
On 5/17/2010 4:54 AM, Sven Schreiber wrote:
Data Analytics Corp. schrieb:
> Hi,
>
> Suppose I have a data set of 200 cars where the observation labels are
> the make (Chevy, Honda, BMW, etc.). Say there are 10 makes. I want to
> create a variable, say ID, that identifies the make so I don't have to
> rely on the observation label. The purpose is to later create dummies
> for the makes to include in a regression model and create boxplots (one
> for each make's price, all shown in one plot window) of price. How can
> I do this? Or is it better to just use the observation labels and
> laboriously write if statements?
>
>
AFAIK the term "observation labels" in gretl terminology is used to
describe individual datapoints, not as a field to hold certain defining
characteristics which is effectively a variable. So, starting from a
spreadsheet file, if you put the "make" variable in its own column (with
character string values) then gretl should automatically code it with
numerical values and notify you when you import that file (in .csv
form). Then it seems to me that you will have what you need: You can
restrict the sample to those obs where the ID equals a certain number
code. If you want you can also create dummies like so:
series hondadummy = (ID=5) # example with code 5 for Honda
This can also be done for all makes in one step in the GUI I think: try
Add -> create dummies for selected discrete variables.
HTH,
sven
--
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt(a)dataanalyticscorp.com
www.dataanalyticscorp.com
12:26 p.m.
Data Analytics Corp. schrieb:
Hi Sven,
I did have make in its own column, the first. I'll try to put it in
another column position and see what happens. If gretl assigns numerics
to the makes, how do I know what numeric goes with what make?
IIRC gretl would write out a text file with the mapping. This fact would
probably also be mentioned in the message gretl gives you on import.
Apart from that it should in principle not make a difference in which
column that variable is stored. But you're right it cannot hurt to try.
The column should not be called 'obs' (or 'date') though, otherwise
gretl tries to infer the dataset structure (e.g. time series) from it.
-sven
5086
days inactive
5089
days old
4 comments
3 participants
participants (3)
-
Data Analytics Corp. -
Dr RJF Hudson -
Sven Schreiber