Quantcast
Channel: EViews
Viewing all 24109 articles
Browse latest View live
↧

Forecast evaluation statistics

April 11, 2018, 6:59 am
$
0
0
Hello Gareth, thanks for your answer!

Assuming that a is the actual series, and f the forecast series, the definition you posted is the same I was using.

However, I didn't manage to replicate the U2 computed by Eviews (I'm using Eviews 10).

I've made a small program to illustrate it, it's in the end of the post.

The scalar "U2", calculated as you told me, is 0.943587.

And the "Theil U2" presented in the forecast evaluation results is 0.971317.

Can you please check if I've made some mistake?

Thanks! Best,
Sergio

Code:

wfcreate m 2013M12 2014M12 
series a
series f
a.fill 1.66, 1.93, 1.95, 2.12, 2.48, 2.62, 2.35, 2.16, 2.12, 1.73, 1.45, 1.24, 0.62
f.fill   NA, 1.84, 1.92, 1.96, 2.31, 3.61, 2.45, 2.01, 1.99, 1.81, 1.65, 1.21, 1.13

genr sq_fpe = ((f-a)/a(-1))^2
genr sq_ape = ((a(-1)-a)/a(-1))^2
scalar fpe2 = @sum(sq_fpe)
scalar ape2 = @sum(sq_ape)
scalar U2 = @sqrt(fpe2/ape2)

a.fcasteval f
show U2


↧
Search

Forecast evaluation statistics

April 11, 2018, 7:46 am
$
0
0
The statistic is only calculated over periods for which both A and F are available both today and yesterday.
Code:

wfcreate m 2013M12 2014M12 
series a
series f
a.fill 1.66, 1.93, 1.95, 2.12, 2.48, 2.62, 2.35, 2.16, 2.12, 1.73, 1.45, 1.24, 0.62
f.fill   NA, 1.84, 1.92, 1.96, 2.31, 3.61, 2.45, 2.01, 1.99, 1.81, 1.65, 1.21, 1.13

series fpe = (f-a)/a(-1)
series ape = (a(-1)-a)/a(-1)
genr sq_fpe = ((f-a)/a(-1))^2
genr sq_ape = ((a(-1)-a)/a(-1))^2
smpl 2014m2 @last
scalar fpe2 = @sum(sq_fpe)
scalar ape2 = @sum(sq_ape)
scalar U2 = @sqrt(fpe2/ape2)
smpl @all
a.fcasteval f
show U2


↧
↧

Add factors

April 11, 2018, 8:35 am
$
0
0
I got the message: Overflow in "A31 equation" in "__KVARTSMODELLEN.ADDINIT(S=B, V=N) @ALL". What does it mean?

The A31 equation is "0*A31+... = ...+0.94748*A31+..." where the 0*A31 has been inserted on the left hand side of the equation (which was transported from a program system where the endogenous variable can be anywhere in the equation) in order to satisfy EViews requirement. That works fine for solving the model without add factors. Moving +0.94748*A31 from the right hand side to replace 0*A31 first on the left hand side removes the error message. Why?

↧

Add factors

April 11, 2018, 8:42 am
$
0
0
Overflow is caused by the model generating a number that is too large for the computer to handle. Usually due to logs (or rather inverting the log to exponentials).

↧

Add factors

April 11, 2018, 8:45 am
$
0
0
Why did the error message disappear when I rearranged the Equation as described in my first email?

↧
↧

Add factors

April 11, 2018, 9:00 am
↧

Add factors

April 11, 2018, 9:29 am
$
0
0
Hi Gareth

When moving our macro model from TROLL to EViews we have to make EViews understand which variable that is to be treated as endogenous in each equation. For example, if we have the equation
y = 4*x + .2*y

and x is to be treated as the endogenous variable, EViews Chris reccommended that we just rewrite the equation as
0*x + y = 4*x + .2*y

see the post http://forums.eviews.com/viewtopic.php?f=10&t=9601&p=33688&hilit=troll#p33688.

Now it seems that this advice creates some problems with respect to initializations of add factors. To see this, run the commands
Code:
wfcreate u 100
    genr x=50+0.03*@trend+2*nrnd
    genr log(y)=3*log(x)+nrnd

model _m
    _m.append 0*x+log(y)=3*log(x)
    '_m.replace  3*log(x)=log(y)
    _m.addassign(v) @all 
    _m.addinit(v=n) @all


Which generates the error message:
Overflow in "X equation" in "_M.ADDINIT(V=N) @ALL".

If you however uncomment the line '_m.replace 3*log(x)=log(y), the addinit command works fine. This is what Dag illustrated in the example above. Is this something you can fix? If so, it would be highly appreciated.
Thomas

↧

Add factors

April 11, 2018, 10:16 am
$
0
0
I just ran that program 1,000 times and it didn't generate an error once...

↧
Search

Concept of SSR of a VAR

April 11, 2018, 10:26 am
$
0
0
I have many VARs and I want to fin d he VAR with the minium SSR.

↧
↧

Forecast evaluation statistics

April 11, 2018, 10:50 am
$
0
0
Thanks Gareth!

I understand the requirement that A is available both today and yesterday, since both the definitions of fpe and ape depend on a(-1).

However, should it be mandatory that F is available both today and yesterday? No definitions depend on f(-1), and when this requirement is included, the one-step ahead forecast error will always be excluded from the computation of U2.

Thanks! Best,
Sergio

↧

Threshold Structural VAR

April 11, 2018, 11:05 am
$
0
0
Hi!

How do I modify threshold value in such a way that lower regime is recession and upper for growth?

Also, what should be the p values of sup, avg and exp wald?

Thank you!

↧

Add factors

April 11, 2018, 11:43 am
$
0
0
What kind of superpowers do you have?

The first picture is the error I get when running the program. I get the second error when I have the model
Code:
_m.append x+log(y)=3*log(x)+x

i.e., I change y=f(x) to x+y=f(x)+x, instead of 0*x+y=f(x).

I have EViews 10, Apr 3. 2018.

It is very strange that you cannot reproduce this... I have also attached a workfile. Try to run the command 
Code:
_m.addinit(v=n) @all
and you should get the error message in the second picture.
Thomas

Error1.PNG

Error2.PNG


↧

Concept of SSR of a VAR

April 11, 2018, 11:58 am
$
0
0
What I'm hinting at is that there are other available metrics for selecting the "best" VAR. If you look at the bottom of a VAR's estimation output you'll find a variety of summary statistics for the VAR as a whole.

↧
↧

Add factors

April 11, 2018, 12:40 pm
$
0
0
Yes, I get the error message on the posted workfile, but if I changed the models specification to:
Code:

0*x + log(y) = 3*log(x)

the error goes away.

↧

Add factors

April 11, 2018, 12:53 pm
$
0
0
that is strange,

do you get an error if you run the command 
Code:
    _m.addinit(v=n) @all

on the attached workfile? In this case the model is: 0 * x + log(y) = 3 * log(x).

When I run it, I get:
1.PNG


Thomas

↧
Search

Add factors

April 11, 2018, 1:08 pm
$
0
0
No error.

Try deleting (or moving so you can move it back) your EViews ini file.

↧

Generate a sum vector

April 11, 2018, 2:12 pm
$
0
0
Hello, I have run this:

for %c {%grp}
for !j = 1 to var_{%c}.@neqn
vector(!j) ssr_{%c}(!j) = var_{%c}.@ssr(!j)
     vector sum_ssr_{%c}(1)=@sum(ssr_{%c})
next
next

where %grp = h dev fixed flex open closed hdebt ldebt

Is there a way to generate a vector like this: ssrt_debt(1)= sum_ssr_ldebt+sum_ssr_hdebt, to sum only the sum_ssr vector for hdebt and for ldebt? I have run this but it doesn't work.

Thank you!

↧
↧

Generate a sum vector

April 11, 2018, 3:07 pm
$
0
0
Hello,

Code:

vector ssrt_debt = sum_ssr_ldebt + sum_ssr_hdebt

That said, if your sum_ssr_{%c} vectors are only ever going to hold one element, use scalars instead of vectors. And there's no point to calculating sum_ssr_{%c} until after you've finished calculating ssr_{%c}, i.e., move that statement after the inner for loop. Finally, you don't need the "(1)" that's currently in your program.

↧

Generate series by sort

April 11, 2018, 6:54 pm
$
0
0
Hello

I'm using IFLS data that contains more than 15000 household data and every household has some person. I want to get the oldest age of person in every household. My friend teach me to use stata with syntax:

Code:
 Bysort hhid14: egen maxage=max(age)


With hhid14 is household id
maxage is series to be generated (maximum age of every household)
age is the age data of every person

But i'm not a stata user. I'm eviews user. And i can't find the bysort command in eviews.

How can i generate series like that? Or there is same bysort command in eviews with different name or word?

↧

Generate series by sort

April 11, 2018, 7:40 pm
$
0
0
Look at @maxsby in the help system.

↧
Viewing all 24109 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>