SAS vs. R... round two: FACTORIAL DESIGN WITH MULTIPLE TREATMENTS

That's right, folks, the great statistical champions are battling it out again in "SAS vs. R: Round 2!" (or, for extra intensity, "round ii"). Today I would like to discuss... FACTORIAL DESIGN WITH MULTIPLE TREATMENTS and SAS's great item for this, PROC MIXED. What is PROC MIXED? PROC MIXED is a SAS command that allows us to do, essentially, a ton of partial regressions within a multilinear regression in order to test statistics about the mean and the variance. What does R have to offer for this same process? I am not sure...

But let me tell you a bit about the SAS. Suppose, for example, we wanted to look at individual means and variances for four treatments blocked on one factor in a RCBD. SAS would say:

PROC MIXED;

CLASS TREATMENT;

MODEL RESPONSE VAR= TREATMENT / DDFM=SATTERTH OUTP=SUMMARY;

LSMEANS TREATMENT / CL;

PROC UNIVARIATE NORMAL PLOT DATA = SUMMARY;

VAR RESID;

PROC GLM DATA=SUMMARY;

CLASS TREATMENT;

MODEL RESID = TREATMENT;

MEANS TREATMENT / HOVTEST= LEVENE(TYPE=ABS);

RUN;

Notice that there is no mention of said means, variances, or amounts of treatment factors in the above SAS. That is because of the "magic" of PROC MIXED. PROC MIXED is able to use the class statements to create the treatment variables. BEST OF ALL... if you have interaction, PROC MIX can be instructed to create variables for the interaction, and test their significance...this is AWESOME, because even with 3 treatments and 3 blocks (1 factor) you are looking at 3^3 possiblities. Do we REALLY want to write all those variables?

I am not sure what R has to offer to combat this... R? Anyone?

I admit that discovery of PROC MIXED has reduced homework time from several hours down to about 30 minutes. Ah.. sigh...

(By the way, every time I have to write "mu" <-- the Greek letter < "mu2" I can't stop laughing. "mu" is not less than "mu2," at least, according to a certain video game that shall not be *pokemon* mentioned.

In Honor of Deliciousness

I couldn't think of anywhere else to show this off; this particular place has, hands down, the best fish jerky I have ever tasted. And... now it can be ordered online. Very pricey, I will probably never order it, but it's cool to know that if, for some reason, one day, if I really, really, really wanted to have Lou's... I could.

http://www.lousfish.com/index.htm

On a side note, a distant relative called my folks today to tell them that he traced our heritage back behind the 1800's and found that our ancestors actually emigrated (I think that is the right word... you emigrate to a place? Uhm....) to Switzerland from Norway in the 1700's... which means... technically... I'm a VIKING.

Long Runs and a Review of "sugar-free Peeps"

For the past few days I have succeeded at a goal... to run 70+ minutes per day. This is how my running used to be, and I remembered that even at a low effort, this duration of run always "worked for me." Gosh, it feels so right to run long again. I think often that I should follow conventional wisdom-- short, fast runs. This, however, doesn't work for me. Maybe in the light of having extra training (martial arts for example) around it, this would work, since the MA would provide the long duration moderate intensity cardio, but without MA, long cardio (with some calisthenics/weights at the end) just feels... right. I am posting this here so that I will remember it... the satisfaction that comes from running long.

This morning I went upwards of 90 minutes. The reason being that I forgot my watch, so I had to rely on my CD player, and the mix list I had which lasts for about 94 minutes. Subtract a few potty breaks in the woods, and that puts me right around 90. Sweetness.

When I was done, of course, I was hungry, but I had no where to park at the school and all my stuff was in the car, so I decided to get a "snack" at Walmart (I needed a binder for something, and one of Clemson's many disadvantages is that there are no office supply stores or Target's here, so Anderson (around 15 miles away) is the closest place to get a binder)... it wasn't worth the drive. I will confess I have a love for marshmellows. They are my bad food weakness. Well, those and vanilla ice cream, although I have a bite of it and I remember that I hate ice cream. It's a cycle with that one. But not with marshmellows. So on sale at Walmart were... sugar-free peeps? I figured, well, they're on sale, and marshmellows aren't exactly healthy anyway, and I just did a run...

My review? They taste exactly like the sugared ones, except my teeth don't feel like they are going to rot out. They have about the same amount of carbs, but it is all from isomalt, which is not any healthier than regular sugar... i.e. they aren't made from Splenda or anything fake so I guess that's why they don't taste like butt... they just aren't technically made from "sugar." I also enjoyed a nice 6 pack of Fresca, since I was really thirsty, and it was near the front of the store, and it reminded me of my childhood.

Seriously, every time I drink Fresca I think of when I was little, and we had a season pass to the zoo. We went every weekend to see the lemurs, and there was a soda machine by the lemurs, so I always got two frescas. That was the same year that Christy's parents took us sailing a lot, and her mom had this big outdoor fridge filled with Fresca and Tab. We used to fill up a cooler with Fresca and sail until our skin cracked. So Fresca is a happy memory for me.

Back to SAS....

Delayed in DFW

I am currently in Dallas, amongst many hats. My flight has been delayed and I am enjoying the strangeness of blogging from a stainless steel keyboard in a very public setting. Unfortunately for anyone who is watching this, nothing I post is particularly exciting. I drank many coffees last night and today, watched a few episodes of NCIS, and tried to rectify the probability distribution thing using Fisher's ... my idea was maybe to split the probability of repeatedly selecting the same colored ball in one distribution (single sample with replacement) from the probability of choosing a red ball from the two red and blue balls (a constant probability of 1/2, but contingent on the selection in the first sample), therefore you have two distributions, but one is dependent on the other. I thought maybe I could use fishers to look at this like two parts of an experiment. The overall probability is different than the sum or the multiple of its parts. Unfortunately, fisher's assumes that the two tests are independent. I need to look further into what to do if they are dependent on one another.

That is all for now. Time to head to SF to see my favorite person.

Shameless OR promotion.

Shameless Promotion of the "Beaver State." In honor of the state that has been mentioned 2 x in the previous posts, I now present:

The amazing moss/crutose/foliose lichen on every tree.

Beautiful rugby fields behind the school, coastal mtns in background, fabulously overcast skies every day!!

Main building by the Beaver Stadium.

Just a little photo evidence.

Lemons -> lemonade?

(Trying to keep this as vague as possible, since we're not supposed to share details of this with anyone outside our lab yet -- so not mentioning the name of my advisor, the name of my current school, etc.)

This morning, my advisor met with our lab, and told us that he's thinking of taking a position at a school in New York. I can't blame him for considering it -- from the way he's described it, he'll have enormously increased freedom to pursue different kinds of research than he does in his current position, not to mention a lot more money to work with, period. He came in this morning, looking apologetic and somewhat defensive, and laid out what this could mean for his graduate students and postdocs, and emphasized that this was definitely not set in stone yet; he may very well end up turning the offer down and staying here.

I have mixed feelings about this. First, the bad part: my girlfriend currently lives in South Carolina, and is moving to Oregon in May -- which is still 9 hours away from the Bay Area, but it's much, much closer! For me to up and move to Long Island would be the ultimate fail, relationship-wise. Another, minor bad part is that I love the weather and the mountains in the Bay Area, and it would suck to move away. I should make the standard noises about how very much I'd miss all my dear friends and social contacts here, but being the terrible recluse that I am, those are pretty much nonexistent.

However, there are a number of positive aspects to this. The main one is that the school in New York actually has real physics and mathematics departments, unlike my current school! Since my research is really more like applied mathematics than biophysics, I think it would be invaluable to have access to a math department where I could discuss, for example, tensor decompositions with people who might actually know something about them, rather than spending months developing my own from scratch, because I work in an intellectual vacuum. Another plus is that because my advisor wouldn't be anchored strictly to NIH money, he'd be free to support research, like mine, that is more speculative, and not directly tied to biomedical applications.

After the rather awkward group meeting this morning, I stopped by my advisor's office, and told him all this, then asked him if it would be possible for me to do most of my work remotely. I pointed out that, although I live only 15 minutes from my current campus, I still do the majority of my work from my apartment. That's one advantage of theoretical research -- you really just need a pen, a pad of paper, and possibly a computer, and you're good to go! So I told him about the situation with my girlfriend, and asked him, "Would it be possible for me to live in Oregon with her, and fly to New York when needed?"

He looked at me oddly, and started to say, "Well, that's, um, that might be a problem..."

"I can fly for free," I added. "My brother works for an airline. So that wouldn't be an issue."

He stared at me for a minute, then laughed. "Oh! So...you're just asking whether it's ok with me if you're not physically at the campus? It doesn't matter to me at all. You're never here anyway. As long as you're making progress, you can live wherever you want."

So, this might not happen at all. But I am thinking that this could turn out to be a really good thing, if it does happen -- I could get a place with my girlfriend in Oregon, periodically fly to New York to a place that actually has physicists and mathematicians to interact with, and at the end of it all, still have a degree with my current school's name stamped on it!

Win?

Today it is 60 in Corvallis

Today it is 60 (and rainy) in Corvallis.

It (was) 16 and windy in Clemson.

I will be glad to move.

:)

Wow, a store that is actually cool

So yesterday I realized... dang, I do not own any pants. Now, this is not troubling for many females, who enjoy shopping, and also who can actually fit INTO pants. But for those of us who have the dimensions of a dude, buying pants is ridiclous and terrible, and elastic waistbands as in track pants are heavenly. Feminine pants look dumb on Germans. Okay, we just don't do "girly."

But... oh, thank you NCIS, I came up with a brilliant idea... what pants are practical, made for people who are muscular, and, best of all... waterproof/resistant and windproof/resistant? Yes, my friends, canvas pants. And where can canvas pants be found but... a store that is actually really cool, the military supply store in Greenville!

I trekked to Greenville to check this place out, half on a mission to find pants and half because I wanted to see all the cool weaponry, artifacts, and gear. It was really neat. I also found pants. They sit up higher than most pants I am used to, but they are really warm and comfortable; I could probably even run in them, if I wanted to. I can deal with it. Best of all, they come in the size I actually wear: unisex small. I have worn this size for roughly the past 15 years, so I figure it's not going to change any time soon.

But what was super cool, and I will have to go back and see, is that they have replicates (and official) gear from all kinds of real battles. You know, real knives, hiking stuff, boots, etc. It was very cool. Flags, you name it.

I hate shopping, but I'll admit, seeing all the hardcore outdoor gear that is actually used in practical situations: that was pretty neat. Maybe I've found a store I actually don't mind going into!

... on a side note: it was 20 degrees and 20 mph winds here today. I forced myself to run 8 miles and then did an intense workout in el gimasio FIKE to prepare for defeating San Bruno. My creepy German genetics did their best to embarass all males at the gym by making me ridiculously strong. It's kind of cool to be strangely strong... but I feel strange because people look at me like "are you really going to lift that?" Yes.

This sucks

My lower back's been bothering me a bit the past few days. Last night it hurt more. This morning I woke up and could barely walk. I tried to go for a run and made it, oh, about twenty steps.

Chiropractor's appointment on Wednesday. Painkillers to last me until then...

Intense pain for the LOSE.

The downside of almonds... and trying to capture a run on camera fail.

So, I was just about to plug in my camera... when I realized that my USB cable appeared to be problematic. Turns out, one of my favorite snacks-- dates and almonds-- had managed to find its way into my USB port... and so I spent the past few minutes picking almonds out of the USB with a random pair of nail clippers that for one reason or another have ended up on my desk (well, with the "file" extension.. I don't think I have ever "filed" anything that wasn't made of wood before in my life, so at least I got to put that thing to good use.)

So why, you ask, was I connecting to my computer? Because this morning I attempted <-- key word -- to simulate San Bruno on the Clemson dikes, with photos to prove it. I have learned a few valuable lessons:

1. I require a crap ton of protein per day to function well in running, and when this amount is off for even a day or so, it makes my running go to poo

2. Lifting weights is apparently a very good workout, as I was ridiculously sore this morning and

3. When there has been an ice storm for 3 days, the ground is very... very... mushy.

Nonetheless. I made it through 6 repeats of the dike. On the seventh I ended up collapsing, which made me think the workout was sufficient. I am attempting to extract the photos from the almond master camera right now... I guess USB cables do not like healthy fats as much as I do!

This is a giant snowball looking up at the side of the dike, which looks completely not intimidating... BUT IT IS!!! It is very intimidating!

This is why I am happy that soon I will be in Oregon. Although admittedly it is kind of pleasant looking.

Just a little 50 minute warm up to get me going!... I love my distance runs.... even in the light of a good workout, it saddens me not to do a distance run, so... heck... might as well do it.

And, if this works, here is the precursor video to the running of the dike. When I succeed at getting 10 repeats of dike (again, only 6 today) I will also make a success film. And then I will work my way up to 20.

Ok. Wow. Back to figuring out some weird definition stuff for stats. I prefer that the math be hard and the text be easy, but in this case, the math is easy and the text is hard. I really shouldn't have been an English major!!

10,000 hours of physics

I recently finished reading Malcolm Gladwell's Outliers, in which he advances the idea that, regardless of your innate talents, in order to achieve true expertise at any complex task, 10,000 hours of practice are required. It's an interesting argument, and the various studies he discusses seem to support this idea forcefully. I got to thinking -- how many hours have I spent purposefully doing at least college-level physics, with the intent to improve my skill at it?

I remember the first time I really put in a lot of time learning physics was for the second semester of the introductory physics courses at UGA, the one that covered basic electricity and magnetism, and optics. That stuff was hard when I first learned it; it's not like the first semester, which covers mechanics, which could be tackled in a pretty common-sense way without spending a lot of time on it. I probably spent about 15 hours a week just on that one class -- so over a semester, something like 240 hours. Then there was linear algebra, which probably was about the same -- 240 hours total for a semester. When I actually started my physics degree, I switched over to taking a full load of physics courses every semester. A full load of physics courses, not surprisingly, is pretty time-intensive. I'd say I averaged at least 40 or 50 hours per week. Including the one semester I spent taking a bunch of random engineering classes at Georgia Tech (which was essentially the same skill-set), that's 5 semesters total -- three at UGA, and two at Tech -- so, assuming 45 hours per week, roughly 3600 hours.

My first year at UCSF was a nightmarish amount of work, but a significant chunk of that time was spent learning random biological material (in particular, structural biology), and learning practical lab skills, neither of which is really the same skill-set as physics, in my mind. So although I was easily spending 80 hours a week studying and/or doing research, I'd say 20 of those, maximum, count in the tally. Probably more in the spring, less in the winter. If it's 20 average, then about 960 for my first year. Second year I TA'd, and had to spend an abnormal amount of time studying statistical physics, then I switched labs to my current lab, where I spend essentially all my time doing physics. So, probably 40 hours per week during the academic year, so about 1440 total. That summer, I spent almost literally every waking hour obsessing over a new tensor method I was trying to develop to analyze a model for interacting magnetic particles (the Ising model) -- I easily totaled 80 hours per week, if not more, over about 3 months, so 960 hours total. Since then, I've settled into a more normal routine; probably around 40 hours per week -- around another 960 hours.

So, tallying it all up: 240 + 240 + 3600 + 960 + 1440 + 960 + 960 = 8400 hours total.

Getting close to the magic number! At this rate, I should hit the 10,000 hour mark exactly 10 months from now -- so, mid-December of this year. Sweet!

Should Foresters Believe in Logs?

The logarithmic transformation is used frequently in forestry calculations. The reason for this is obvious, and true: we deal with LOGS.

Specifically, we deal with the volume of logs, and therefore, the value of logs.

Many foresters assume that a log is cylindrical. Well... logs are somewhat cylindrical, but ask anyone who has been to a pole mill: a good pole tree is hard to find. In other words, it is rare that a perfectly cylindrical tree is found, or even a perfectly cylindrical log cut from a tree. Trees taper off at their upper ends. In fact, we measure tree height in what is called merchantable height, which means "a certain minimum width at the top"-- usually 4-6 in. The bole of the tree could be 22 inches... somewhere up high, it will be 4-6 inches.

Let's say we're walking through the woods and we want to know "how much is this tree worth for timber?" You can look at a 70 foot pine with a 15 inch DBH, but that means that 16 feet up that tree you aren't looking at a 15 inch DBH-- you may only have a 12 inch DBH. So how do we measure the tree? Let's simply consider the case of a perfectly straight, but tapered tree. First of all, we've got to make an estimate of the tree's diameter at some height above DBH in all of these methods-- and let me tell you right now: no economical forester is going to go climb every tree to get this measurement. When you've got to survey 200 acres in a day, you learn to eyeball it, or, if you're lucky, maybe get a good laser gator. Next, you've got to figure out what cuts you are going to make-- you can get 8, 10, 12 and 16 foot logs at most places, plus crappy wood like rail tie and pulpwood. These cuts represent the "height" of your cylinder. Usually at this point you will cut the tree. Then you choose a log rule: Doyle's, Scribner's, or International-- you can make an arguement for any of the three rules (each has advantages and disadvantages) to determine the merchantable volume from the log, and hope you get a good estimate....

It seems like, in theory, foresters should believe in "logs." That is, if the volume of a tree is ROUGHLY cylindrical, we can use the cylinder volume equation: (pi/4)*(d^2)*h in a logarithimic form: log(volume) = log (pi/4) + 2 log (d) + log(height) which could convert to log(volume) = b0 + b1D + b2H where D = log(d) and H = log(h) to estimate the proper equation given lots of data about diameter and height... but we forget something magical and terrible about logs: where there is a logarithmic transformation, there is also the ability for the data variance to be increased and the distribution to be non-normal. And over here, we do not do non-normal distributions. Well... we try not to!

Increase variance: decrease ability of the equation to work for large values of d and h... and who wants an equation that only works accurately on small, crappy trees...

So I wonder... what's a forester to do? The answer many mills use is actually.. well, calculus. New laser scanners can actually integrate the shape of a log and provide an accurate measurement of a cut logs volume-- they can even suggest optimal cuts to get the best grain and least knots. Got to love technology, but I admit: it seems awfully ironic that foresters really can't believe in LOGS.

The Cloak

So what did I do that was useful today... not much.

For starters, I got blizzarded into the MSP airport this morning, where I worked with tree math for a while and then I watched an episode of NCIS that I downloaded yesterday called "The Cloak." No, I am not going to give a synopsis here, but I think this episode had a pretty cool point. So I do have to give a synopsis.

The idea of the episode was that the main characters were engaged in a war game with another group; they were looking for a security breech. And their superiors were engaged in another war game above them looking for the security breech. And there were several levels of this game going on, and the whole show ended up with the eventual way of dealing with the breech simply being to follow the "mole" until it went back to its bosses. This, I think, mimics how I SHOULD deal with data. I like to invent "games" for data on many levels. Maybe I should just try to trace it some time, instead of wondering how it plays...

By the way: the "cloak" as mentioned above is cool, but snuggies are not.

Gregor Mendel LP

Okay, so this is really dumb, and the video is worthless, but the song is HILARIOUS

Want to make fun of the fundamentals of genetics?

Hi, Meiosis

Yes, how exciting. I work hard over here.

Couldn't sleep, but wow, heard a cool song!

I get major test anxiety and man, I could not sleep last night at all. However, I think it must be my new regime of protein and banana consumption that is giving me a butt load of energy, because at 4:15 this morning, for the first time in a WHILE (since summer camp, really) I woke up just wanting to run-- it was 31 out and kind of damp-- excellent!

While I was awake last night, I searched for this cool song that I had only heard a few words of. I have no idea why I like it so much, but it really is catchy-- and it's a great running tune. I had it in my head the whole run; it's got a slow start, but really, it's pretty great. Never heard of this group before, "Blue October," but I've not heard of many groups!

http://popup.lala.com/popup/432627069328786377

Stuff that can't be both

I have always been really interested in endocrinology.

I was talking about o. chem. yesterday in the office (because when graduate students get off topic, it's just switching to another area of academia, I think), but I really amazed by how many hormones have this "either one or the other, but not both" thing going on... ex. seretonin and norepineprhine; insulin and cortisol; adrenaline; insulin and glucagon; leptin and glucagon, etc. And then there's the whole limitation on hormonal creation because of the fundamental molecule-- ex. DHEA in relation to adrenaline and cortisol-- too much adrenaline is actually treated by increasing DHEA... even though adrenaline is synthesized from DHEA. Type 2 diabetes: too much insulin is treated with insulin injections-- under utilization treated effectively by supplementation. It defies "common logic" but it's in the fundamental nature of things.

What really makes me curious about this is that we look at many biological things as if they are going to have this asymptotic, "normal" function; I guess I'm thinking about marginal cost (thanks, Pericles! :) )or something here, but we might logically surmise, that the creation, for example, of too much insulin, will mean that each additional insulin will have a decreasing effectiveness... I guess the logic is that they would have an equal effect and just be bonding to a decreasing amount of available insulin receptors. The fact is, it's more discrete that that-- too much insulin, and you inhibit the functions of cortisol and glucagon making the effect of insulin-- more effective...

I learned that there are receptors in chloroplasts sensitive to wavelengths of red light-- if you have too many trees growing near one another, one of the reasons for distressed growth isn't a lack of light, but the fact that light reflected off of other trees is actually green and not red. So that the red light receptors are turned off, and the tree is not only harmed by the lack of direct red light from the sides, but from the additional effect of having its red light receptors turned off...

As a martial artist, I remember the best fighting tactic I learned was: whatever seems natural, do the opposite-- if you are being choked, move into the choke. If someone is punching you, step in past the punch and take a forearm to his throat, if someone tries to pull you down, fall. The oddest thing about this to me was that although typical fighting techniques seemed logical, the "atypical" fighting techniques actually agreed with the qigong/meridian theory-- it was as if we naturally have an atypical nature, but logically have a typical one---

I wonder if this is sort of a governing principle of all things: endocrinology, fighting, math, whatever: we are naturally inclined to be "one" or "the other" but not both-- trying to be marginally in "two worlds" results in failure to be in either one. Evolution is crazy like that, I guess. Maybe that's something philosophical there, too. I don't know. I guess I'll read this later and think either, "wow, that's dumb," or "wow, that's brilliant." And there's another duality for you: I don't think I'll read it and say, "well, that was about 40% okay."

I guess the long and short of it is that some stuff just can't be both.

Other people's poop on my bathroom floor = bad day

The bad news: yesterday, my toilet overflowed. It was disgusting. I mucked out the bathroom and plunged the toilet for about a half hour, but it didn't seem to unclog. I had a big meeting the next morning, so I went to bed and figured I'd deal with it the next day. This morning, the toilet had overflowed AGAIN, and this time, a bunch of rancid water had come up my shower drain. Absolutely disgusting. Then I got back from work, and I noticed that the main drain outside my apartment was overflowing with what looked like raw sewage. About an hour later, the public works emergency crew was there, and told me that yes, this actually was a sewage overflow, and yes, this was raw sewage strewn across my building's parking area...and my apartment's bathroom. That's nice to know that the rancid liquid I was mucking out was, in fact, raw, untreated sewage.

The good news: they fixed it, and the landlord (an unpleasant character who I've previously locked horns with...a little over a year ago, we had a screaming match over the phone, where he threatened to take me to court because I had the temerity to replace my apartment's heater which was sparking all over my carpet and an obvious fire hazard, after he ignored two written notices from me over the course of two and a half months...we hadn't been on speaking terms since then) came by, actually apologetic for once. He had the maintenance guy come into my apartment and clean out my bathroom, which was nice, so I don't have to be exposed again to a bunch of raw sewage. He also gave me $200 off next month's rent, so it's not a total loss!

Matrix Algebra: 1, Calculus: 0

I am reviewing my least squares in preparation for this exam on Thursday and I saw this on Wikipedia. For the first time, I thought to myself, wow... I understand the matrices more than the calculus!

Moss...

Today I spend the day being a Bryologist.

This morning I decided to run up in the CEF. It was 26 and very icey out, which made for a precarious run, but it was really fun. I was a muddy mess by the time I was done; the thawing ice was making rivers and a lot of trails were closed. As I ran out on the trails (I did some single track) I noticed a lot of moss-- as I ran back, none. Why? Moss grows on certain (I believe north-facing) aspects.... interesting.

This afternoon we took the ecology class to a big granite rock to teach them about succession. I got to pull up a bunch of Spagnum spp. (Reindeer moss) which is one of my favorites. We also saw a ton of Vaccineum arboreum (Tree Sparkleberry) just growing its way to happiness. I've always liked Tree Sparkleberry... it looks a lot like Ligustrum sinense (Privet), which I don't like at all, but Tree Sparkleberry is the "native" version.

It was good to see some Tree Sparkleberry. I looked for running shoes today, as well, but no DS Trainers in site. Guess I'll be wearing my flats until I can find some!

Going to CA in 4 days!!! Yes!!