Tuesday, April 1, 2008

Blogging competitions and Bayesian Kriging

It’s day one of the blogging competition. I have been thinking about how to manage this so that it will be most interesting for you and me both. I’m sure you’re not interested in an hourly log of my daily activities (and neither do I care to write you one). I could tell you stories about my day, which is more or less the model I’ve used so far on my blog. I could write extensively and primarily about the optional “topic of the day” (part of the contest), but I can already tell you I will not have 3 let alone 300 words to say about some of the topics. For instance, if I'm not mistaken, today's topic of the day is "fonts," as in Times New Roman, etc. I don't know anything about fonts. I knew a guy in college, however, who was a Graphic Design major and he had an entire semester course on just fonts (!!) so apparently it is a large topic. Later, after college, I was looking at job openings at Hallmark and discovered that the person who writes the messages on greeting cards is not the same person who picks the font and designs the presentation of the message. Those are two separate positions. And neither of these is the person who comes up with the card concept in the first place. Except for a few trivia about the first movable type printing presses, that's really all I know about fonts. So clearly, some days I would find it difficult to write substantially on the topic of the day.

Alternatively, I thought about having themed days on one or two given days of the week—kind of like families that always, invariably have pizza on Saturday night. I could instate “Switzerland Saturdays,” for instance, and bring you an interesting weekly column relating to my country of residence (which would be appropriate on a blog that claims to be about my adventures here anyway). Other theme days options include sharing information from my classes and projects (Epidemiology Tuesday, or African Literature Wednesdays, or Refugee Thursdays). So, I am open to taking suggestions for regular weekly “column” days (I’ll do up to two a week), but they have to relate to something I’m doing and/or interested in. So don’t even bother suggesting stuff like baseball, economics, and underwater basket weaving. But anything that you think I might know something about (ha! That limits it right there!) and that you would like to know more about is fair game. You can post them on here, or tell me in some other way. Ok ‘nuff said.

Last night I went to my epidemiology lecture on the spatial patterns of infant mortality in Tanzania. Now, if you ever wondered how you would use a degree in statistics, epidemiology is a good application. For instance, yesterday’s presentation was a Master student’s project. The research team connected with the Swiss Tropical Institute took available information on infant mortality rates and demographic and socio-economic factors in certain limited clusters throughout the country and combined this info with environmental factors (soil types, distance from bodies of water, annual rainfall, etc.). They used Bayesian Geostatistical models to assess the infant mortality risk based on this information and then applied something called Bayesian kriging (more statistical stuff) to estimate infant mortality in the areas for which they don’t actually have data and to produce a smooth map showing the areas that have the highest risk. Governments and other health care organizations can use this information to focus their interventions on the areas that need it most in order to curb infant mortality. Interesting, hmm? I don’t understand Bayesian kriging one bit, but it looks like an interesting topic to read up on for you math-minded people. Here are some links:

www-users.york.ac.uk/~pml1/bayes/preface.htm

www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/Gebhardt+Gebhardt.pdf

www.stat.washington.edu/peter/spatial/Covnew.pdf

3 comments:

  1. Well, maybe I have this thing figured out-- we will see whether it allows me to post.

    It will be a delight to see all three bazillion of the posts you will enter in order to win your contest, but I have serious doubts as to your ability to maintain the schedule...especially as you are planning som eoutings, travel, etc. in the middle of it all.

    You set up your poll in such a way that I could not give an honest answer-- I would have voted, no, I don't think you have time to do this. I do NOT think it is dumb or boring. (Although I think a trip to a car museum is most probably not high on your to-do list...)

    BTW, I liked the pastel drawing.

    Dad

    ReplyDelete
  2. Hello my most favoite Papa! I'm writing my blog entries on the tram and on my 7 minute breaks in classes, so I'm making good use of "dead" time. Normally car museums aren't a priority, it's true, but I might go just cuz Gregg would be disappointed if I didn't. glad you liked the drawing. I'm planning another one for Thursday afternoon so stay tuned for more artistic endeavors. :)

    ReplyDelete
  3. i think an epidemiology on the epidermis would be fascinating. i have a strange obsession with doing Google image searches for skin diseases. your writing might be more satisfying than just pictures. also, nice to meet you fellow competitor.

    ReplyDelete