Hi all,
For problem 2c, there seem to be four values that need to be set before
implementing the function we defined in part 2b: lambda, y, alpha, beta. We
are told alpha and beta. How do we know what lambda and y are defined to be?
Thanks,
Jacqueline
Hi All,
Just a quick note/reminder that tomorrow following lecture (e.g. 4 PM)
until 6:30 PM, Ian and I will be hosting an optional R help session in
N-018. Please bring questions/frustrations/challenges you've faced, and
we'll try to bring some challenging tasks as well... And if you come
late, that's fine; we'll be set up to handle that.
Best,
Dan
----
Ph.D. Student
Department of Government
Harvard University
Tutor, Currier House
dhopkins(a)fas.harvard.edu
http://www.danhopkins.org
The below is a good question and worth responding to via the list. There
are definitely a variety of parameterizations for the Weibull, and indeed
the variant we present in the problem set is *not* the same as the
parameterization in R. The variant in the problem set should, however,
make the underlying math relatively transparent (for those who are
curious). Best, Dan
On Sat, 4 Mar 2006, Daniel Fetter wrote:
> hi guys,
>
> is the density function for the weibull given on the problem set the
> right one? it doesn't correspond to that given in R or on the few
> websites i found. it seems like it should be
>
> f(x|alpha,beta)=beta*((alpha*x)^(beta-1))*exp(-(alpha*x)^beta)), rather than
> f(x|alpha,beta)=beta*alpha*(x^(beta-1))*exp(-alpha*(x^beta))
>
> (this is using alpha as the reciprocal of R's scale parameter b).
>
> thanks.
> dan
>
>
> On 3/4/06, Dan Hopkins <dhopkins(a)fas.harvard.edu> wrote:
> > Hi All,
> >
> > With many thanks to some of you guys for passing along these errors, we've
> > identified and corrected two errors in problem set 3. A corrected version
> > is posted here:
> >
> > http://www.courses.fas.harvard.edu/~gov2001/Assignments/pset3.pdf
> >
> > In 2(a), the final sentence referred to a "joint probability," but since
> > there is a continuous parameter, it should be referred to a "probability
> > density." (The function only produces traditional probability statements
> > after integrating over the continuous parameter.)
> >
> > In 4(b), the sigma_1,2 covariance parameter was supposed to be 0.6, not
> > 2.4.
> >
> > Also, the video from section 4 is now posted. And Ian confirms that the
> > mvtnorm library is available on the icegov machines, so you do *not* need
> > to install it.
> >
> > Good weekends to all...
> >
> > Best,
> > Dan
> >
> >
> >
> > ----
> > Ph.D. Student
> > Department of Government
> > Harvard University
> > Tutor, Currier House
> > dhopkins(a)fas.harvard.edu
> > http://www.danhopkins.org
> > _______________________________________________
> > gov2001-l mailing list
> > gov2001-l(a)lists.fas.harvard.edu
> > http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
> >
>
Hi All,
With many thanks to some of you guys for passing along these errors, we've
identified and corrected two errors in problem set 3. A corrected version
is posted here:
http://www.courses.fas.harvard.edu/~gov2001/Assignments/pset3.pdf
In 2(a), the final sentence referred to a "joint probability," but since
there is a continuous parameter, it should be referred to a "probability
density." (The function only produces traditional probability statements
after integrating over the continuous parameter.)
In 4(b), the sigma_1,2 covariance parameter was supposed to be 0.6, not
2.4.
Also, the video from section 4 is now posted. And Ian confirms that the
mvtnorm library is available on the icegov machines, so you do *not* need
to install it.
Good weekends to all...
Best,
Dan
----
Ph.D. Student
Department of Government
Harvard University
Tutor, Currier House
dhopkins(a)fas.harvard.edu
http://www.danhopkins.org
Hi All,
In an effort to keep your inboxes from overflowing, we've consolidated a
good deal of useful information into this email.
PROBLEM SETS
1) Ian has drafted a very helpful page on numerical integration which
should make the homework a bit easier. You can find it here:
http://www.courses.fas.harvard.edu/~gov2001/sections/numint.pdf
***As you'll notice, although we use logs to improve numerical stability
when calculating the joint density, the ultimate joint density we want is
not on the log scale--and so the final line of code on page 1 puts us back
onto the regular scale by exponentiating the logged joint density--that
is, taking the exp() of the logged joint density. *This is an important
point we neglected to mention in section.*
2) Homework 3 is posted on the website and due next Thursday, March 9th,
in section. Just a reminder that except for distance students, *we expect
hard copies turned in either at section or to Dan's mailbox on the 3rd
floor of CGIS before section.*
SECTIONS
3) Many thanks to those who were at section for the feedback provided
through the notecards. One key recommendation was to focus more on the
specific tasks and functions needed for the homework. We will very much
keep this in mind as we plan for future sections.
4) In a course with people from varying backgrounds, ongoing feedback is
especially key, so Ian and I encourage people to email us individually or
leave notes in my mailbox--they don't have to be signed--with any feedback
you might have.
5) Based on strong demand, we will go ahead with an R Help session this
Monday, March 6th after lecture (4 PM) until at least 6:30 PM in CGIS room
N-018. We will try to generate some exercises, but if you plan to come,
come with questions/frustrations you want to address.
6) We've uploaded the section 4 notes to the website (designed to help
with the homework), as well as the "name that distribution" pdf that we
discussed briefly in the 6 PM section and at more length in the 5 PM
section. We've also uploaded the code needed to make those figures in the
.pdf, so you can experiment with the probability density functions and
cumulative density functions.
7) Email list--After the discussion in the 6 PM section about overflowing
inboxes, I wanted to point out that you can set the Gov. 2001 list options
via this webpage:
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
Good weekends to all.
Best,
Dan
----
Ph.D. Student
Department of Government
Harvard University
Tutor, Currier House
dhopkins(a)fas.harvard.edu
http://www.danhopkins.org
Our discussion of first principles in section tonight made me wonder, "Why
don't we have a big table of distributions with their first principles, PDF,
CDF, etc.?" Further, why don't the terms used on Wikipeda match the terms
we use in class?
Why don't we take apart this Wiki on the beta distribution
http://en.wikipedia.org/wiki/Beta_distribution and rebuild it using
GOVT-2001 terminology.
Signed,
Aiming without a Reference
I'm out of the office from Friday 3/3/06 to Monday 3/6/06 and will be back
in the afternoon of 3/6/06.
I will respond to your message when I return.
Have a good weekend!
Lynda
Hi All,
The question below is relevant to all. We are interested in assessing the
probabilities so that you can decide whether to switch or not to. The
probabilities that we care about are those relevant to that decision...
Best,
Dan
On Wed, 22 Feb 2006, Delia Bailey wrote:
> Hi Dan and Ian,
> Just a quick question about PS2...
> for the monty hall problem, in the instance where you pick the
> winning door AND monty gives you the prize, do you want us to lump
> that together with "win, no switch" since you don't get a chance to
> switch? or calculate the percentage of time that it's best to switch,
> conditional on monty not giving you the prize immediately?
> does that make sense?
> -delia
>