Dear Riccardo, thanks for your answer. I manage to export from the database a csv file with the same content I report in the previous mail. So now I can open the file in Gretl. Now the difficult part is to transform the database in the adequate way in order to obtain a panel data set. Thanks for the suggestions about how to design the data set. We know we are dealing with a difficult conceptual task. I think I will need aditional dummys, one for each supermarket in order to capture specific effects.
But still, I need to work around the issue of creating the panel. Any other suggestion is welcome. Thanks Leandro 2010/4/3 Riccardo (Jack) Lucchetti <r.lucchetti(a)univpm.it> > On Sat, 3 Apr 2010, Leandro Zipitria wrote: > > Hello Gretl Community, >> >> I have a database with daily prices which is in a rather unusual format >> and I want to know if it is possibly to create a panel data with it using >> Gretl scripts. The database is >> extracted from a dbf archive, and its has 6 variables: >> >> - The (number of) supermarket from which the price is reported >> - The (number of) the product which price is being reported >> - The year of the price >> - The month of the price >> - The price reported itself >> - The first day of the month the price is reported >> - The last day of the month the price is reported >> >> I have nearly one million rows with data, but in order to do some >> regressions -and use it as a panel- I will need to transform it in a >> suitable way. >> >> I suppose that the best way that Gretl can handle it is to create a >> specific column for each product, and then stack all the supermarkets on a >> daily basis. In this way, I will >> have each column representing a product, the first 700 rows being a price >> for each product for supermarket 1, the next 700 rows being a price for each >> product for supermarket >> 2, etc. >> But in order to do it, I will need first to create the daily prices >> series, which is now "compacted" in the datafile. I am attaching a random 10 >> elements from the database in >> order to get a better picture of the situation. In the first sheet I >> submit the actual data format, on the second one which I think should be the >> (best?) result. >> >> I will first ask if this kind of transformation is possible in Gretl. I am >> aware that running some scripts on other programs could do the trick, but I >> think that it could be >> possible in Gretl to do it. But I am also think that it could be rather >> complex to do it, and I am a new one on this issues. >> > > Two remarks/suggestions: > > 1) Turning a dataset such as this into a panel dataset is not trivial from > a _conceptual_ point of view: what are your units? supermarkets or products? > or combinations of the two? What would you use as the time unit (day, month, > week)? If you choose anything longer than a day (say, a week), how would you > handle changes of prices during the week? Of course, this is a design > decision and gretl can't help you with this. > > 2) If you have your data in some database that can be queried via SQL, I > think that our ODBC apparatus may just fit your needs. You may want to have > a look at the corresponding chapter of the User's Guide. > > > Riccardo (Jack) Lucchetti > Dipartimento di Economia > Università Politecnica delle Marche > > r.lucchetti(a)univpm.it > http://www.econ.univpm.it/lucchetti > _______________________________________________ > Gretl-users mailing list > Gretl-users(a)lists.wfu.edu > http://lists.wfu.edu/mailman/listinfo/gretl-users >Dear Riccardo,
thanks for your answer. I manage to export from the database a csv file with the same content I report in the previous mail. So now I can open the file in Gretl. Now the difficult part is to transform the database in the adequate way in order to obtain a panel data set.
Thanks for the suggestions about how to design the data set. We know we are dealing with a difficult conceptual task. I think I will need aditional dummys, one for each supermarket in order to capture specific effects.
But still, I need to work around the issue of creating the panel.
Any other suggestion is welcome.
Thanks
Leandro
2010/4/3 Riccardo (Jack) Lucchetti <r.lucche...@univpm.it>
Two remarks/suggestions:On Sat, 3 Apr 2010, Leandro Zipitria wrote:
Hello Gretl Community,
I have a database with daily prices which is in a rather unusual format and I want to know if it is possibly to create a panel data with it using Gretl scripts. The database is
extracted from a dbf archive, and its has 6 variables:
- The (number of) supermarket from which the price is reported
- The (number of) the product which price is being reported
- The year of the price
- The month of the price
- The price reported itself
- The first day of the month the price is reported
- The last day of the month the price is reported
I have nearly one million rows with data, but in order to do some regressions -and use it as a panel- I will need to transform it in a suitable way.
I suppose that the best way that Gretl can handle it is to create a specific column for each product, and then stack all the supermarkets on a daily basis. In this way, I will
have each column representing a product, the first 700 rows being a price for each product for supermarket 1, the next 700 rows being a price for each product for supermarket
2, etc.
But in order to do it, I will need first to create the daily prices series, which is now "compacted" in the datafile. I am attaching a random 10 elements from the database in
order to get a better picture of the situation. In the first sheet I submit the actual data format, on the second one which I think should be the (best?) result.
I will first ask if this kind of transformation is possible in Gretl. I am aware that running some scripts on other programs could do the trick, but I think that it could be
possible in Gretl to do it. But I am also think that it could be rather complex to do it, and I am a new one on this issues.
1) Turning a dataset such as this into a panel dataset is not trivial from a _conceptual_ point of view: what are your units? supermarkets or products? or combinations of the two? What would you use as the time unit (day, month, week)? If you choose anything longer than a day (say, a week), how would you handle changes of prices during the week? Of course, this is a design decision and gretl can't help you with this.
2) If you have your data in some database that can be queried via SQL, I think that our ODBC apparatus may just fit your needs. You may want to have a look at the corresponding chapter of the User's Guide.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucche...@univpm.it
http://www.econ.univpm.it/lucchetti
_______________________________________________
Gretl-users mailing list
gretl-us...@lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
