2:54 p.m.
Hi,
In the following code (as a simplified example), how do I retrieve the name
of the series Y so as the OLS output shows the actual name of the variable
instead of just "Y"? I use the "varnames()" for the X-list variables
names but couldn't come out with something feasible for Y
<\hansl
function matrix m (series Y , list X )
ols Y X
end function
<hansl>
Best,
Artur
3:37 p.m.
On Thu, 31 Dec 2020, Artur Bala wrote:
Hi,
In the following code (as a simplified example), how do I retrieve the name
of the series Y so as the OLS output shows the actual name of the variable
instead of just "Y"? I use the "varnames()" for the X-list
variables
names but couldn't come out with something feasible for Y
<\hansl
function matrix m (series Y , list X )
ols Y X
end function
<hansl>
Use the argname() function.
Allin
8:47 a.m.
Use the argname() function.
Allin
Yeah, that's quite clear but argname() generates a string not a
series. And my point is different. Let's have the following:
<\hans>
function matrix myfunction (series depY, list X)
# some coding
ols depY X
# some coding
return something
end function
<hans>
Now, when I run my function, let's say "mymatrix = myfunction(Price,
xlist)", OLS printing will show up: "Dependent variable: depY" instead
of "Dependent variable: Price".
Is there a way to have the actual name of depY displayed?
Best,
Artur
10:32 a.m.
Am 04.01.2021 um 09:47 schrieb Artur Bala:
> Use the argname() function.
>
> Allin
Yeah, that's quite clear but argname() generates a string not a
series. And my point is different. Let's have the following:
<\hans>
function matrix myfunction (series depY, list X)
# some coding
ols depY X
# some coding
return something
end function
<hans>
Now, when I run my function, let's say "mymatrix = myfunction(Price,
xlist)", OLS printing will show up: "Dependent variable: depY" instead
of "Dependent variable: Price".
Is there a way to have the actual name of depY displayed?
You could copy the input using argname, for example:
ID = genseries(argname(depY), depY)
Then you should be able to use the ID number also in the OLS command.
Of course the copying doesn't make it 100% efficient, but the loss
should be minor.
cheers
sven
10:37 a.m.
On Mon, 4 Jan 2021, Sven Schreiber wrote:
> Now, when I run my function, let's say "mymatrix =
myfunction(Price,
> xlist)", OLS printing will show up: "Dependent variable: depY"
instead
> of "Dependent variable: Price".
> Is there a way to have the actual name of depY displayed?
You could copy the input using argname, for example:
ID = genseries(argname(depY), depY)
Sorry, a question to Artur: of course, Sven solution is perfectly ok,
but... why would you need to do this?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
12:09 p.m.
Le lun. 4 janv. 2021 à 11:37, Riccardo (Jack) Lucchetti <
p002264(a)staff.univpm.it> a écrit :
On Mon, 4 Jan 2021, Sven Schreiber wrote:
>> Now, when I run my function, let's say "mymatrix = myfunction(Price,
>> xlist)", OLS printing will show up: "Dependent variable: depY"
instead
>> of "Dependent variable: Price".
>> Is there a way to have the actual name of depY displayed?
> You could copy the input using argname, for example:
> ID = genseries(argname(depY), depY
Thank you Sven!
Sorry, a question to Artur: of course, Sven solution is perfectly ok,
but... why would you need to do this?
Oops, not sure I could convincingly argue for this :) but my feeling is
that when one enters a séries name in a function he/she expects to get back
the same name to référ to. Probably the use of '*' may be a solution to
that...
Best,
Artur
12:20 p.m.
Am 04.01.2021 um 13:09 schrieb Artur Bala:
> Sorry, a question to Artur: of course, Sven solution
is perfectly ok,
> but... why would you need to do this?
> Oops, not sure I could
convincingly argue for this :) but my feeling is
> that when one enters a séries name in a function he/she expects to get
> back the same name to référ to. Probably the use of '*' may be a
> solution to that...
No, the pointer form doesn't change the name inside the function. In
fact, how could you write your function if you didn't know under what
identifier the input could be referenced? That's the whole point of
modularized code, that you don't have to care about the names in the
outside world.
But I do understand what you mean. In a different context gretl has a
mechanism to change the display name, namely through 'setinfo <series --graph-name=<displaystring>'. I guess you want to
have something like
that for the printout in models. The problem is (I think) that so much
code would have to be changed for all the possible estimation commands.
So why not just copy the series and be done...
cheers
sven
12:45 p.m.
Le lun. 4 janv. 2021 à 13:20, Sven Schreiber <svetosch(a)gmx.net> a écrit :
Right :) !
Am 04.01.2021 um 13:09 schrieb Artur Bala:
No, the pointer form doesn't change the name inside the function.
Good to know!
In
fact, how could you write your function if you didn't know under
what
identifier the input could be referenced?
It does make sense. My understanding was that refering to *y as input will
bring into the function all the attributes of y. One just would have to
refer to "y" as identifier eventually.
But I do understand what you mean. In a different context gretl has a
mechanism to change the display name, namely through 'setinfo
<series
--graph-name=<displaystring>'. I guess you want to have something like
that for the printout in models.
Right!
So why not just copy the series and be done...
1:22 p.m.
On Mon, 4 Jan 2021, Sven Schreiber wrote:
Am 04.01.2021 um 09:47 schrieb Artur Bala:
>> Use the argname() function.
>>
>> Allin
>
> Yeah, that's quite clear but argname() generates a string not a
> series. And my point is different. Let's have the following:
> <\hans>
> function matrix myfunction (series depY, list X)
> # some coding
> ols depY X
> # some coding
> return something
> end function
> <hans>
>
> Now, when I run my function, let's say "mymatrix = myfunction(Price,
> xlist)", OLS printing will show up: "Dependent variable: depY"
instead
> of "Dependent variable: Price".
> Is there a way to have the actual name of depY displayed?
You could copy the input using argname, for example:
ID = genseries(argname(depY), depY)
Then you should be able to use the ID number also in the OLS command.
Of course the copying doesn't make it 100% efficient, but the loss
should be minor.
Or if you don't mind using string substitution, you could use
"rename":
<hansl>
function matrix myfunction (series depY, list X)
string realname = argname(depY)
rename depY @realname
ols @realname X
...
end function
</hansl>
Allin
1:34 p.m.
On Mon, 4 Jan 2021, Allin Cottrell wrote:
On Mon, 4 Jan 2021, Sven Schreiber wrote:
> Am 04.01.2021 um 09:47 schrieb Artur Bala:
>>> Use the argname() function.
>>>
>>> Allin
>>
>> Yeah, that's quite clear but argname() generates a string not a
>> series. And my point is different. Let's have the following:
>> <\hans>
>> function matrix myfunction (series depY, list X)
>> # some coding
>> ols depY X
>> # some coding
>> return something
>> end function
>> <hans>
>>
>> Now, when I run my function, let's say "mymatrix = myfunction(Price,
>> xlist)", OLS printing will show up: "Dependent variable: depY"
instead
>> of "Dependent variable: Price".
>> Is there a way to have the actual name of depY displayed?
>
> You could copy the input using argname, for example:
> ID = genseries(argname(depY), depY)
>
> Then you should be able to use the ID number also in the OLS command.
> Of course the copying doesn't make it 100% efficient, but the loss
> should be minor.
Or if you don't mind using string substitution, you could use "rename":
<hansl>
function matrix myfunction (series depY, list X)
string realname = argname(depY)
rename depY @realname
ols @realname X
...
end function
</hansl>
The problem with these approaches is that what you pass to the function
may not be a valid identifier, for example
myfunction(log(y), X)
and anything based on argname would break down in that case. If what you
want is be sure that what your function prints makes sense I would suggest
to fine-tune the printout using tools like printf, sprintf() and
modprint, which give you complete control on what to print and how. For
example, here's a completley silly function that returns the OLS
coefficient, halved.
<hansl>
set verbose off
function matrix OLS_half(const series y, list X)
sname = argname(y)
if strlen(sname) == 0
sname = "(anonymous series)"
endif
parnames = varnames(X)
ols y X --quiet
bs = ($coeff ~ $stderr) / 2
printf "Silly output for %s\n", sname
modprint bs parnames
return bs
end function
open data4-1.gdt
list X = const sqft
ols price X
OLS_half(price, X)
OLS_half(log(price), X)
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
2:04 p.m.
On Mon, 4 Jan 2021, Riccardo (Jack) Lucchetti wrote:
On Mon, 4 Jan 2021, Allin Cottrell wrote:
> On Mon, 4 Jan 2021, Sven Schreiber wrote:
>
>> Am 04.01.2021 um 09:47 schrieb Artur Bala:
>>>> Use the argname() function.
>>>>
>>>> Allin
>>>
>>> Yeah, that's quite clear but argname() generates a string not a
>>> series. And my point is different. Let's have the following:
>>> <\hans>
>>> function matrix myfunction (series depY, list X)
>>> # some coding
>>> ols depY X
>>> # some coding
>>> return something
>>> end function
>>> <hans>
>>>
>>> Now, when I run my function, let's say "mymatrix =
myfunction(Price,
>>> xlist)", OLS printing will show up: "Dependent variable: depY"
instead
>>> of "Dependent variable: Price".
>>> Is there a way to have the actual name of depY displayed?
>>
>> You could copy the input using argname, for example:
>> ID = genseries(argname(depY), depY)
>>
>> Then you should be able to use the ID number also in the OLS command.
>> Of course the copying doesn't make it 100% efficient, but the loss
>> should be minor.
>
> Or if you don't mind using string substitution, you could use
"rename":
>
> <hansl>
> function matrix myfunction (series depY, list X)
> string realname = argname(depY)
> rename depY @realname
> ols @realname X
> ...
> end function
> </hansl>
The problem with these approaches is that what you pass to the function may
not be a valid identifier, for example
myfunction(log(y), X)
and anything based on argname would break down in that case.
Good point, but the optional argument to argname can be used to
prevent breakage:
string realname = argname(depY, "depY")
If what you want is be sure that what your function prints makes
sense I would suggest to fine-tune the printout using tools like
printf, sprintf() and modprint, which give you complete control on
what to print and how. [...]
Agreed, that's the clean solution.
Allin
2:29 p.m.
Good point, but the optional argument to argname can be used to
prevent breakage:
string realname = argname(depY, "depY")
> If what you want is be sure that what your function prints makes
> sense I would suggest to fine-tune the printout using tools like
> printf, sprintf() and modprint, which give you complete control on
> what to print and how. [...]
Agreed, that's the clean solution.
Yes, I just rediscovered the 'modprint'...definitely, a valuable tool !!!
Thank you all for your time and inputs,
Artur
1:57 p.m.
Or if you don't mind using string substitution, you could use
"rename":
<hansl>
function matrix myfunction (series depY, list X)
string realname = argname(depY)
rename depY @realname
ols @realname X
...
end function
</hansl>
That's cool too. Indeed I tried something similar but missed to add "@" for
'realname'...and obviously, it didn't work :)
Thank you Allin,
Artur
1202
days inactive
1206
days old
12 comments
4 participants
participants (4)
-
Allin Cottrell -
Artur Bala -
Riccardo (Jack) Lucchetti -
Sven Schreiber