PDA

View Full Version : Regression Final

Bühlmann
01-10-2006, 01:58 PM
Word on the street was that today's exam was kinda hard. Wonder what that means for the Time Series final?

Selena
01-10-2006, 02:25 PM

Selena
01-10-2006, 02:33 PM
Does anyone know the formula for elastisities (sp?) ? I thought it was B (Beta), the slope coefficient?

jingle47
01-10-2006, 02:35 PM

Bühlmann
01-10-2006, 02:41 PM

I'm sitting for Time Series. A co-worker took Regression.

bucks22
01-10-2006, 03:27 PM
I think there were a few typos on the exam as well. There is nothing more frustrating than that when taking a test. Did anyone else notice this?

joeorez
01-10-2006, 03:28 PM
Elasticity = (Beta Hat j) * (X Bar j / Y Bar).

Anyone get the wrong answer for RSS in question 6, and therefore unable to continue with 7 through 13? Not a good start.

jingle47
01-10-2006, 03:50 PM
How about the ARCH question? I totally skipped the reading and so I guessed A for #49

Selena
01-10-2006, 03:57 PM
Elasticity = (Beta Hat j) * (X Bar j / Y Bar).

Anyone get the wrong answer for RSS in question 6, and therefore unable to continue with 7 through 13? Not a good start.

I multiplied my answer by 100 to get their answer and continued.

NewTubaBoy
01-10-2006, 04:57 PM
People thought it was hard... crap... I guess I should go and do some studying for my time series final. The LAST thing I want to do is fail this exam and have to retake the darn final.

Selena
01-10-2006, 05:26 PM

I will send you my answers if you send me yours. PM me

NewTubaBoy
01-10-2006, 05:39 PM
typos aside, for those who actually took it, do you also agree that it was more difficult than either the practice problems or than expected?

joeorez
01-10-2006, 06:49 PM
Which problems do you think had typos?

pi_rsquare
01-11-2006, 01:22 AM
I multiplied my answer by 100 to get their answer and continued.

Yep, I did exactly the same thing :)

red
01-11-2006, 10:57 AM
Can anyone confirm that it was harder than the sample problems that were posted? I took Regression in July and the test was the same as the samples. I am getting ready to take Time Series on Friday and want to know if I should be putting in a lot more time.

Thanks.

jingle47
01-11-2006, 11:07 AM
I would put a little more time into it....

Patriots' Fan
01-11-2006, 11:47 AM
Elasticity = (Beta Hat j) * (X Bar j / Y Bar).

Anyone get the wrong answer for RSS in question 6, and therefore unable to continue with 7 through 13? Not a good start.

Did anyone get something like 0.36 for the variance? It was from there on I wasn't sure if I was systematically wrong because I wasn't sure about what I got for the variance... does 0.36 sound familiar to anyone?

Selena
01-11-2006, 12:54 PM
Did anyone get something like 0.36 for the variance? It was from there on I wasn't sure if I was systematically wrong because I wasn't sure about what I got for the variance... does 0.36 sound familiar to anyone?

Yes I got 0.0036, that answer was not on their choice list, so I multiplied it by 100 and chose 0.36 as an answer. I figured that 0.36 is 0.0036 x 100%, since we were dealing with %.

Patriots' Fan
01-11-2006, 01:15 PM
Yes I got 0.0036, that answer was not on their choice list, so I multiplied it by 100 and chose 0.36 as an answer. I figured that 0.36 is 0.0036 x 100%, since we were dealing with %.

And then the answer for one of the questions, maybe RSS, was like +1?

Selena
01-11-2006, 02:03 PM
And then the answer for one of the questions, maybe RSS, was like +1?

I do not think so

Patriots' Fan
01-11-2006, 02:17 PM
I do not think so

Or was it TSS, or R^2? what did you get for them?

Selena
01-11-2006, 02:40 PM
Or was it TSS, or R^2? what did you get for them?

I do not remember now. TSS, ESS, RSS were all <1, but something else was +1...do not remember what it was thou.

joeorez
01-11-2006, 03:15 PM
I am curious which questions people think had typos.

jingle47
01-11-2006, 03:36 PM
did anyone else notice that two of the choices for the F-test were negative?
That narrows things down a bit.

Also:on # 17, Choices A,B,and C gave that alpha would go up, down, or stay the same.....So you know it has to be 1 of those 3 and therefore can eliminate choice D or E no matter what the question asked.

cincinnatikid
01-11-2006, 04:41 PM
Also:on # 17, Choices A,B,and C gave that alpha would go up, down, or stay the same.....So you know it has to be 1 of those 3 and therefore can eliminate choice D or E no matter what the question asked.

Seems logical, and I struggled but I actually think I chose either D or E on that one....

IIRC, Choices A, B, and C had to do with the change in alpha if the X's were increased by 1. If you increase X by 1, the change in alpha depends on whether beta is positive or negative, so I threw out all of A, B, and C. Adding 1 to the Y's appears to ALWAYS (perhaps that word should have been added to the choices) increase alpha by 1, so I picked whichever choice that was.

Patriots' Fan
01-11-2006, 05:47 PM
Wasn't the beta negative?

cincinnatikid
01-12-2006, 10:45 AM
Wasn't the beta negative?

I didn't get the impression it was relating to any particular alpha or beta, but that the question was about the relationship in more of a general sense.

Chubbs
01-12-2006, 11:10 AM
I am curious which questions people think had typos.

Question 21 had a typo. In the question they tell you that B (hat) is .500 and then ask you to talk about the change in B hat. But all the answers relate to B hat being greater to or less than .600

This typo wasn't too hard to figure out, I think they meant to use .500 in all the answers, but reading the question the way it was written, you can't solve it.

Also, in question 28 they ask for the elasticity of the variable X2.

Partial deriv of Y with respect to X2 is .4
X2 bar is -.2
Y bar is .08

.4 * -.2/.08 is -1 I assumed they made a decimal place error and chose -.1

cincinnatikid
01-12-2006, 11:13 AM
Question 21 had a typo. In the question they tell you that B (hat) is .500 and then ask you to talk about the change in B hat. But all the answers relate to B hat being greater to or less than .600

This typo wasn't too hard to figure out, I think they meant to use .500 in all the answers, but reading the question the way it was written, you can't solve it.

Also, in question 28 they ask for the elasticity of the variable X2.

Partial deriv of Y with respect to X2 is .4
X2 bar is -.2
Y bar is .08

.4 * -.2/.08 is -1 I assumed they made a decimal place error and chose -.1

Agree on both of these.

Patriots' Fan
01-12-2006, 11:57 AM
For one of the answers I got something like 0.25 and it only has the choice 0.20. I was so sure about my answer I just chose 0.20 and moved on.

Does anyone remember this problem?

Chubbs
01-12-2006, 01:06 PM
Did anyone get something like 0.36 for the variance? It was from there on I wasn't sure if I was systematically wrong because I wasn't sure about what I got for the variance... does 0.36 sound familiar to anyone?

In question 1.8, they ask for the varience of the regression. Is that what you are talking about. Because I got .0039% and didn't need to multiply anything.

In question 1.9 they ask for the variance of the B estimator. This is the varience of the regression divided by the sum of the (X-Xbar).

This yields a nice clean 3.9% The rub is, you have answers of .39% and 3.89%

I chose the .39% because I was assuming a decimal error, but I didn't multiply by 100, it was only 10. and i was using 39 not 36, so I am little confused as to what you may be talking about

Patriots' Fan
01-12-2006, 01:36 PM
In question 1.8, they ask for the varience of the regression. Is that what you are talking about. Because I got .0039% and didn't need to multiply anything.

In question 1.9 they ask for the variance of the B estimator. This is the varience of the regression divided by the sum of the (X-Xbar).

This yields a nice clean 3.9% The rub is, you have answers of .39% and 3.89%

I chose the .39% because I was assuming a decimal error, but I didn't multiply by 100, it was only 10. and i was using 39 not 36, so I am little confused as to what you may be talking about

How did you get .0039% for the variance of the regression?

Yeah, I got 3.89% too for the variance of beta and picked the 3.9% answer. But I was more thinking of the variance of the regression problem.

Chubbs
01-12-2006, 01:47 PM
How did you get .0039% for the variance of the regression?

Yeah, I got 3.89% too for the variance of beta and picked the 3.9% answer. But I was more thinking of the variance of the regression problem.

The variance of the regression can be found by dividing the ESS by the degrees of freemdom.

.0117/3=.0039

Selena
01-13-2006, 09:18 AM
Also, in question 28 they ask for the elasticity of the variable X2.

Partial deriv of Y with respect to X2 is .4
X2 bar is -.2
Y bar is .08

.4 * -.2/.08 is -1 I assumed they made a decimal place error and chose -.1

Did either of you send an email to NEAS mentioning these errors? If not I will send one.

jingle47
01-13-2006, 11:07 AM
I think you should send an email whether they did or not.