bananas stats course

Replies to This Discussion

Permalink Reply by TheBananaGirl on July 14, 2010 at 1:35pm

Wow this is very exciting Prad! Thanks to you and vegamama for organising this :)

Permalink Reply by The Humane Hominid on July 14, 2010 at 4:32pm

It's been a while since I did stats, so I'll be here. :)

Permalink Reply by pradtf on July 24, 2010 at 7:52am

ok we are just about ready to start!

below are some of the details of the course which will be conducted in its own separate thread with an explanation of both content and process. we will maintain a set pace for it, though people are free to go at whatever speed they wish to.

we will be using the very nicely done carnegie-mellon university free online course as the trunk with various branches growing out of it directly relating to the china study data.

if you want to get a headstart go here:
http://oli.web.cmu.edu/openlearning/forstudents/freecourses/statistics

the software of choice is R which you can get from here:
http://www.r-project.org/
(though if you are on a linux or bsd system, it is likely your distribution has its own install process for it.)
we prefer R because it is open source, gnu software and exceptionally good - and was strongly recommended by rayna!
however, if some of you want to use excel, you can, since the course offers that opportunity too.

if you need help with installation, just ask in this thread.

the course thread location will be announced in another post.

in friendship,
prad


=======
the cm course is done very clearly with plenty of illustrative examples
and interactive practice exercises. the presentation is more practical
than mathematical in orientation, so instead of, say, figuring out
derivation or proofs, you get to actually work with the data right away
- and understand what you are producing.

datasets for exercises can be downloaded and used in a free analysis
program such as R.

there are even explanatory videos!

there are 4 main units after a good introduction to the course:

unit 2: exploratory data analysis
techniques for sumarizing and making sense of the data

module 1 examining distributions
- categorical and quantitative variables

module 2 examining relationships
- 4 types of relations (three used in course cases I, II, III)
- causation with nice job on confounders


unit 3: producing data
deals with sampling methods and different study designs

module 3 sampling
- sampling plans random and non-random

module 4 designing studies
- observational, exeriments, surveys
- role of causation


unit 4: probability
- preparatory groundwork for drawing inferences

module 5: introduction
- basic concepts

module 6: finding probabilities
- frequencies, outcomes, various rules

module 7: conditional probabilities
- conditional and independence, multiplication rule, trees

module 8: random variables
- discrete, continuous

module 9: sampling distributions
- parameters, behaviors


unit 5: inference
- using sample data to draw conclusions

module 10: introduction
- forms of statistical inference

module 11: inference for one variable
- estimations, hypothesis testing

module 12: inference for relationships
- cases I, II, III

Permalink Reply by sparrowrose on July 24, 2010 at 11:39am

"which will be conducted in its own separate thread"

Will you link to that thread from this one to be sure that no one accidentally misses it?

Thank you!

Permalink Reply by pradtf on July 24, 2010 at 11:53am

we certainly will sparrowrose!
it'll likely be up a bit later tonight.

in friendship,
prad

Permalink Reply by pradtf on July 24, 2010 at 3:58pm

banalysis is here!
where?
right here:
http://www.30bananasaday.com/group/debunkingthechinastudycritics/fo...

in friendship,
prad

Permalink Reply by pradtf on July 30, 2010 at 1:34pm

this post is just experimental. we are trying to explore the best way to represent code in a post and may be removed.

in friendship,
prad

======

bold
#identify the url
dataurl <- "http://www.ctsu.ox.ac.uk/~china/monograph/CH83PRU.CSV"

#get the data from the url
chinadata <- read.csv (dataurl, na.strings = ".", strip.white = TRUE)

#then pull out the xiang 3 and sex T
chinadata1 <- subset ( chinadata, Xiang == 3 & Sex == "T")


italic
#identify the url
dataurl <- "http://www.ctsu.ox.ac.uk/~china/monograph/CH83PRU.CSV"

#get the data from the url
chinadata <- read.csv (dataurl, na.strings = ".", strip.white = TRUE)

#then pull out the xiang 3 and sex T
chinadata1 <- subset ( chinadata, Xiang == 3 & Sex == "T")


using the pre tag

#identify the url
dataurl <- "http://www.ctsu.ox.ac.uk/~china/monograph/CH83PRU.CSV"



#get the data from the url

chinadata <- read.csv (dataurl, na.strings = ".", strip.white = TRUE)



#then pull out the xiang 3 and sex T

chinadata1 <- subset ( chinadata, Xiang == 3 & Sex == "T")

Permalink Reply by veganmama18 on July 30, 2010 at 1:39pm

i like the last one (using the pre tag). how did you do that?

Permalink Reply by pradtf on July 30, 2010 at 2:45pm

you can enclose things between <pre> and </pre>

i like it too - but there seem to be serious issues with it on ning
for instance, it is putting extra lines when we don't want them.
also, if your line is too long, it will run off the edge.

i'm inclined to think that bold or italic will likely be best.

you can play with it here with some code you have and see what the output is like - we haven't done more than in the last post because we went for dindin, but based on what we did, i have a feeling it's not going to work well here.

in friendship,
prad