Log in

No account? Create an account
UCLA Biostatistics Ph.D. Students (2001)'s Journal

> recent entries
> calendar
> friends
> UCLA Biostatistics
> profile

Thursday, July 26th, 2001
5:48 pm

Today I did two more previous questions while timing myself. It went pretty smoothly although I'm frustrated with my memory sometimes. Things that I've worked with extensively in the past I can completely forget if I haven't seen them in the past few weeks. That means I'll have to have very condensed materials that I expose myself to just days before the exam -- i.e. cramming -- if I'm to do well. I'm generally good at remembering where to look things up. I guess I've mostly had the luxury of turning to my books to confirm details. Unfortunately, that won't be possible in the exam.

I need to post the solutions to some problems I've worked recently. I also need to finish putting together my "Regression Facts" summary and get going on my "Asymptotics Facts" and "Distribution Facts" and "Inference Facts" ... and whatever other facts I'm going to make. I think I'll do "Survival Facts" too. Ahhh...too much to do. I'm counting down the weeks.

(1 comment | comment on this)

Friday, July 6th, 2001
4:22 pm

Today I reviewed some applied regression and worked a regression problem from the 1997 Ph.D. exam. I also ordered the Mathematical Statistics text that I mentioned from half.com. I'm debating buying Stapleton's Linear Statistical Models. It's my favorite of the advanced regression texts. It's a little pricey though so for now I've just re-requested it via inter-library loan.

current mood: worried

(comment on this)

Thursday, July 5th, 2001
5:54 pm

Today I worked through the first two chapters of Jun Shao's Mathematical Statistics. Two concepts that I've refreshed my understanding of are sufficiency and completeness. I'd like to photocopy problems from the chapters before the book's due back in the library.

I also worked one transformation problem from the 1999 qualifying exam. Tonight I'll post the work on that problem. Thanks Li-Jung for pointing out the problem.

I need to work more problems!

(comment on this)

Monday, July 2nd, 2001
4:23 pm

So today I got a key to the biostat room A1-250. I'm beginning my daily morning study schedule. I'll be studying on campus from 9 a.m. to 2 p.m. everyday.

I began today by doing some review of transformations and moment generating functions in Casella and Berger. I plan to become familiar with how all the most common transformations among the most frequently distributions are done.

I've determined that I should try to work about 1 Ph.D. qualifying exam question per day in parallel with my review. Aside from that I'll hold aside the 2000 exam to try about 1 month before the actual exam to try to gauge where I stand.

Today went pretty well and I feel like I actually have an idea of how to proceed. I worked about 5 excercises today. One was somewhat interesting and I may post it this evening.

I've decided that my succinct notes will likely take on the form of "Important Facts about ..." I think the only topic that I will try to make fairly comprehensive notes about will be survival analysis because it's the topic I feel I'll need to review the most.

(comment on this)

Sunday, June 24th, 2001
12:54 am - First up...

I think I'm starting with probability. I'm trying to get some exposure to the relevant parts of measure theory. I'd like to understand a few key theorems...the dominated convergence theorem and the Radon-Nikodym theorem. It seems like if I were more comfortable with the mathematics behind probability that it would be more accessible.

From there I'm going to go onto transformations and generating functions as part of a review of the most commonly used distributions.

I've really got to make some sort of list and calendar. I'm actually feeling better prepared for the theoretical than the applied sections right now. I need a good review of survival and contingency table methods. There's so much and I need to review virtually all of it. I think my operational approach -- as far as what I actually *do* as I study -- will be to create a condensed set of notes for myself and to create a catalog of worked out problems. It's important for me to actually produce useful "stuff" that I can look at again in the future. If I think that just reading and doing will help me for the exam I'm only slightly right because things reviewed in June will be largely forgotten by September. I need to create really outstanding notes that I can continue to turn to.

(comment on this)

Thursday, June 21st, 2001
2:25 am - Ready to get started?

So the summer has begun and with it, the studying. It feels like I have all the time in the world right now, but I know how the weeks cruise by. Hmmm...well, I guess I need to make my study plan.

Here's a reminder of the quick list of topic

1. Probability, random variables, common distributions (especially those related to the MVN and survival analysis), generating functions, transformations.

2. Modes of convergence and convergence theorems.

3. Inference -- especially MLE and M-estimation

4. Order statistics and quantile function, rank statistics, confidence bands for distribution functions and quantile functions.

5. Linear models, Gauss-Markov theorem, LS estimation, hypothesis testing, multiple comparisons, mixed models, variance components, Bayesian estimation in linear models, resampling techniques.

6. Multivariate analysis, estimation, testing and classification, longitudinal data, growth curves, RCR models, principal components, factor analysis, canonical correlation, discriminant analysis.

Ok. So looking this over, the areas where I feel weakest are confidence bands and survival analysis. Confidence bands because I haven't learned it well and survival because I haven't done it in so long. For the others I feel like I'm well grounded but need to brush up and just get a lot more experience by working problems. Time to hit the books! Ok, well the pillow first, but the books in the morning.

Anyone who wants to join in is welcome.

current mood: determined

(comment on this)

> top of page