7:44 a.m.
Hello to all the data wizards out there,
today I hit the limit in the GUI that the earliest year can be set to
1500. But I was looking at the really historic time series from here:
http://www.ggdc.net/maddison/maddison-project/orihome.htm, which
actually starts at 1 A.D. It worked ok via script, but I think it also
should work via the dialog window.
Now let's see if I manage to load that gappy data into the workfile...
no, there are problems, and I think some of them are bugs. (This is
1.9.90 on Win7.)
When I start with an empty annual dataset from 1 to 2100 and try to
append the Maddison data from an Excel worksheet (where I have named the
year column with "date"), the rows/years are not properly matched
against the inner years ("inner" in the sense from 'join'). That's
because of the (huge) gaps in the source file. Strangely, when I use
"obs" instead of "date" then gretl says instead I must not use this as
a
variable name.
I also have to rename many many variables in the xls file before gretl
accepts them, and I think this is really not the optimal way to handle
this because it's very time-consuming and dull; there should be some
automagic "mangling" of the names by gretl, maybe accompanied by a
warning message, or the whole mangling could be a user-configurable option.
Then I tried to treat the whole thing as a (country) panel structure --
but I'm noticing (for the first time although it must have been there
for ages) that when I choose "new dataset" from the menu, the dialog
forces on me the detour to specify the overall number of obs (anybody
got a calculator ready?) and then afterwards only can I impose the panel
structure. Suggestion: why not have radio buttons with
cross-section/time-series/panel in that dialog, and in the panel case
let the user input numbers for both dimensions right away. (plus the
periodicity for time series and panel as well)
Another suggestion: why not allow the use of a time index variable for
time series the same way that index variables are allowed for panels?
I haven't succeeded so far with the import, the only solution I can
think of right now are to add hundreds of empty rows to the source file
to remove the gaps.
Hm.
cheers,
sven
8:56 p.m.
On Mon, 26 May 2014, Sven Schreiber wrote:
today I hit the limit in the GUI that the earliest year can be set
to 1500.
OK, that ought to be made more flexible.
But I was looking at the really historic time series from here:
http://www.ggdc.net/maddison/maddison-project/orihome.htm, which
actually starts at 1 A.D. It worked ok via script, but I think it also
should work via the dialog window.
Now let's see if I manage to load that gappy data into the
workfile... no, there are problems, and I think some of them are
bugs. (This is 1.9.90 on Win7.)
When I start with an empty annual dataset from 1 to 2100 and try
to append the Maddison data from an Excel worksheet (where I have
named the year column with "date"), the rows/years are not
properly matched against the inner years ("inner" in the sense
from 'join'). That's because of the (huge) gaps in the source
file. Strangely, when I use "obs" instead of "date" then gretl
says instead I must not use this as a variable name.
We're talking about (I think)
http://www.ggdc.net/maddison/maddison-project/data/mpd_2013-01.xlsx
Well, "obs" is a reserved word in gretl while "date" is not.
Normally that wouldn't matter but in this case it does because the
hugely gappy Maddison data are not recognized by gretl as any sort
of time series so gretl is apparently trying to read the first
column as data.
I also have to rename many many variables in the xls file before
gretl accepts them, and I think this is really not the optimal way
to handle this because it's very time-consuming and dull; there
should be some automagic "mangling" of the names by gretl [...]
I see that Stata has infinite patience with silly-buggers column
headings and in CVS I've made a move in that direction. But I need
to rant just a little: I suppose the data are carefully done, but
the column headings look as if they've been added by a total
computer illiterate with the attitude "I don't care; if anyone
really wants to make sense of this they'll find a way".
Then I tried to treat the whole thing as a (country) panel
structure -- but I'm noticing (for the first time although it must
have been there for ages) that when I choose "new dataset" from
the menu, the dialog forces on me the detour to specify the
overall number of obs (anybody got a calculator ready?)
Yes, on the gretl toolbar ;-)
and then afterwards only can I impose the panel structure [...]
Another suggestion: why not allow the use of a time index variable
for time series the same way that index variables are allowed for
panels?
Maybe, but this is not needed for any ordinary time series.
I haven't succeeded so far with the import, the only solution I
can think of right now are to add hundreds of empty rows to the
source file to remove the gaps.
Importing these data is not trivial since although they have a time
dimension they do not resemble any ordinary time-series. I have to
wonder what the advantage is in having them recognized as a "time
series" of sorts; I can't imagine that any time-series methods could
be applied to them.
Anyway, you can get them to appear as annual data in gretl as
follows. I began by opening the file in gnumeric and exporting it as
text, with all fields quoted. One could use ssconvert. You then have
to fix a few variable names (and add a couple of missing ones) -- or
use current CVS to have the varnames handled automatically. In
addition I gave the first column a heading of "myear" so that gretl
wouldn't try to interpret it as time. Then,
<hansl>
open maddison.txt -q
set skip_missing off
matrix X = {dataset}
string varnames = ""
nv = $nvars - 1
loop j=1..nv -q
varnames += varname(j)
varnames += " "
endloop
nulldata 2010 --preserve
# annual, starting in 1 AD
setobs 1 1 --time-series
list L = null
loop j=1..nv -q
string vname = strsplit(varnames, j)
series @vname
L += @vname
endloop
loop s=1..rows(X) -q
# year is in first col of X
scalar t = X[s,1]
loop foreach j L -q
$j[t] = X[s,j]
endloop
endloop
</hansl>
It's a bit complicated, but our importers are just not designed to
handle this sort of thing. (For anyone who hasn't seen these data,
they are "sort of" annual time series but with gaps of many
centuries between the earlier observations.)
Allin
1:57 a.m.
Am 27.05.2014 02:56, schrieb Allin Cottrell:
On Mon, 26 May 2014, Sven Schreiber wrote:
Well, "obs" is a reserved word in gretl while "date" is not.
I thought "date" was a synonym for "obs" for the purposes of
importing,
but apparently I mixed things up.
Normally that wouldn't matter but in this case it does because
the
hugely gappy Maddison data are not recognized by gretl as any sort
of time series so gretl is apparently trying to read the first
column as data.
But shouldn't gretl in this case tell the user that it doesn't recognize
the "obs" values as periods, instead of complaining about the header?
> I also have to rename many many variables in the xls file before
> gretl accepts them, and I think this is really not the optimal way
> to handle this because it's very time-consuming and dull; there
> should be some automagic "mangling" of the names by gretl [...]
I see that Stata has infinite patience with silly-buggers column
headings and in CVS I've made a move in that direction. But I need
to rant just a little: I suppose the data are carefully done, but
the column headings look as if they've been added by a total
computer illiterate with the attitude "I don't care; if anyone
really wants to make sense of this they'll find a way".
Rant granted. But OTOH I think one has to concede that with this stuff
Excel sets a kind of standard that people come to expect of other
software as well. It's like blanks in file names; I hate them and try to
educate others to hate them as well, but in practice on Windows one has
to admit that they usually just work.
> Then I tried to treat the whole thing as a (country) panel
> structure -- but I'm noticing (for the first time although it must
> have been there for ages) that when I choose "new dataset" from
> the menu, the dialog forces on me the detour to specify the
> overall number of obs (anybody got a calculator ready?)
Yes, on the gretl toolbar ;-)
Yep, but I still see the potential to roll several dialogs into one there.
> Another suggestion: why not allow the use of a time index
variable
> for time series the same way that index variables are allowed for
> panels?
Maybe, but this is not needed for any ordinary time series.
Right, the need only appeared with these funny series.
Importing these data is not trivial since although they have a time
dimension they do not resemble any ordinary time-series. I have to
wonder what the advantage is in having them recognized as a "time
series" of sorts; I can't imagine that any time-series methods could
be applied to them.
All I wanted to do is to plot them, for example in a bar graph (so that
NAs really don't get plotted or connected). The point is that I use
gretl often only as a data manager for things like transformations or
plots, without doing econometrics on the data, because gretl is usually
very good at it.
In the end I used Excel which I don't like much for plotting, and it
also did not work 100%, but at least I got something.
Anyway, you can get them to appear as annual data in gretl as
follows. I began by opening the file in gnumeric and exporting it as
text, with all fields quoted. One could use ssconvert. You then have
to fix a few variable names (and add a couple of missing ones) -- or
use current CVS to have the varnames handled automatically. In
addition I gave the first column a heading of "myear" so that gretl
wouldn't try to interpret it as time. Then,
...
I really appreciate your efforts, you really don't have to provide
tailor-made solutions to questions like these. When I report these
problems it's to extend gretl in the future, not (usually) to get the
work done for me. Anyway, since I wanted to plot the data, I could have
imported them in a matrix like you did and then use the built-in matrix
plotting feature of the 'gnuplot' command. I'll try this.
Thanks!
Sven
2:48 p.m.
On Tue, 27 May 2014, Sven Schreiber wrote:
Am 27.05.2014 02:56, schrieb Allin Cottrell:
> I see that Stata has infinite patience with silly-buggers column
> headings and in CVS I've made a move in that direction. But I need
> to rant just a little: I suppose the data are carefully done, but
> the column headings look as if they've been added by a total
> computer illiterate with the attitude "I don't care; if anyone
> really wants to make sense of this they'll find a way".
Rant granted. But OTOH I think one has to concede that with this stuff
Excel sets a kind of standard that people come to expect of other
software as well. It's like blanks in file names; I hate them and try to
educate others to hate them as well, but in practice on Windows one has
to admit that they usually just work.
Sigh. But yes, you're right.
> Anyway, you can get [the Maddison data] to appear as annual data
in
> gretl as follows. [...]
I really appreciate your efforts, you really don't have to provide
tailor-made solutions to questions like these. When I report these
problems it's to extend gretl in the future, not (usually) to get the
work done for me.
Thanks, I understand. But I wasn't aiming to "do your work for you" so
much as show that it could be done in hansl, if not exactly conveniently
then without extreme measures such as inserting hundreds of blank rows in
the source file.
Anyway, I've now generalized the more automated handling of bad column
headings to the spreadsheet importers, so this will now work "out of the
box":
open mpd_2013-01.xlsx --rowoffset=2
Since the first column heading is blank the years come in as observation
markers. To get the years as a series one could insert a heading such as
"datayear" in the first column and reopen the data file, or -- showcasing
"introspection" via hansl's string-handling:
<hansl>
markers --to-file=markers.txt
string s = readfile("markers.txt")
matrix yearvec
sscanf(s, "%m", yearvec)
series year = yearvec
</hansl>
Allin
8:05 p.m.
On Wed, 28 May 2014, Sven Schreiber wrote:
Am 27.05.2014 20:48, schrieb Allin Cottrell:
> <hansl>
> markers --to-file=markers.txt
Interesting, I wasn't really aware of 'markers'. According to the docs
you can do --to-file and --from-file; what about displaying/printing them?
Well, they're displayed whenever you print/display one or more
series with --byobs (or via the GUI). But I guess we could let the
"markers" command with no options just print the markers by
themselves (or say "No markers" if there aren't any).
Allin
3824
days inactive
3828
days old
5 comments
2 participants
participants (2)
-
Allin Cottrell -
Sven Schreiber