This our re-do of the dependent sample assignment from 2/10 Driving Reaction Time Under Distractive Conditions.docx
Monday 2/7/2011
Function
|
Action
|
length(object)
|
number of elements/components
|
dim(object)
|
dimensions of an object
|
str(object)
|
structure of an object
|
class(object)
|
class or type of an object
|
mode(object)
|
how an object is stored
|
names(object)
|
names of components in an object
|
c(object, object,...)
|
combines objects into a vector
|
cbind(object, object, ...)
|
combines objects as columns
|
rbind(object, object, ...)
|
combines objects as rows
|
object
|
prints the object
|
head(object)
|
list the first part of the object
|
tail(object)
|
list the last part of the object
|
ls()
|
list current objects
|
rm(object)
|
delete an object
|
newobject <- edit(object)
|
edit object and save as newobject
|
fix(object)
|
edit in place
|
To import data from an EXCEL file - RODBC package to access the files
install.packages ("RODBC")
Webscraping - the user extracts info imbedded in a web page using:
readlines ( ) grep ( ) gsub ( ) functions
Sunday 2/6/2011
Example for Linear Least Squares Regression:
http://www.cyclismo.org/tutorial/R/linearLeastSquares.html
When we type varialbename<-lm(what ever values we pass), we should issue summary(variablename) to get Residuals, Coefficients, Residual standard error, ...etc. issuing variablename will only show the coefficients! plot(myvariable) will give multiple plots (should discuss how to read them!!)
runif(n, min=0, max=1): It let us generate random deviates; however, unlike mvrnorm we can't specify the mean and the covariance matrix. We forgot to ask how we can specify min and max limit so we don't give unwanted values!
useful commands:
help.search("a topic")
example("a topic")
data() shows all datasets in the loaded packages
sink(file = "filename", append = FALSE, type = c("output", "message"), split = FALSE) to write the output to the file filename :) "split = TRUE write to both the file and the screen. Not quite sure what type is?
sink(): return the output back to the screen.
source("myfile") will submit the code from myfile to the current R session.
png("myfile.png") to redirect graphic output to myfile. We have to issue the drawing command (plot() for example) and then dev.off(). we can also use jpeg(), bmp() or pdf().
help(package="packagename"): to list the functions and datasets included in packagename
Shoe Wear Graphics
library(MASS)
data(shoes)
granova.ds(data.frame(shoes),revc=TRUE, main="Shoe Wear",xlab="Sole material B",ylab="Sole material A")
"More about comparing Two independent samples." Assignment:
To create the dataframe: Check http://msenux.redwoods.edu/math/R/dataframe.php for more information.
musicTypeA=c(26,21,22,26,19,22,26,25,24,21,23,23,18,29,22)
musicTypeB=c(18, 23, 21, 20, 20, 29, 20, 16, 20, 26, 21, 25, 17, 18, 19)
Lowry.A.B.dt=data.frame(A=musicTypeA,B=musicTypeB)
The updated function "es.twogps" that includes calculating the CI:
es.twogps <- function(x, y)
{#generates standardized ES for input
# x,y (indep.) sample data
mx <- mean(x)
my <- mean(y)
nx <- length(x)
#nb: nx will generally NOT equal ny
ny <- length(y)
md <- mx - my
vpoold <-((nx-1)*var(x) +(ny-1)*
var(y))/ (nx + ny - 2)
sdpoold <- sqrt(vpoold)
ES <- md/sdpoold
md<-round(md,2)
ES<-round(ES,2)
GM=22
CIL=GM-(2*sdpoold)
list(md=md,ES=ES,CIL=CIL,CIU=CIU)
Questions:
- What is: (in simple linear regression)
- influence statistics
- Residuals standard error
- F-statistic
- Multiple R-squared
- Adjusted R-squared
Comments (5)
bob pruzek said
at 6:56 pm on Feb 6, 2011
I will provide an alternative way, maybe two, for calculating a ( 95% ) confidence interval for the population mean difference. The direct method you've used is not recommended (tho' it might, as we shall see, be close
numerically to what I shall provide) More on this Monday. bp
Dima Kassab said
at 7:33 pm on Feb 6, 2011
Thanks Bob!
bob pruzek said
at 1:05 am on Feb 7, 2011
Dima (or ??), You asked: "Questions:
What is: (in simple linear regression)
influence statistics
Residuals standard error
F-statistic
Multiple R-squared
Adjusted R-squared"
We have not covered regression yet, so these questions will have to wait a bit for detailed answers. But I will bring a book or two that should help in the meantime. And I can provide many URLs that will also help
if you want a head start.As for some of the other things on this page, I'm not sure if there are q's you want me to answer, but if any of you (in the whole class) do, just make the q's explicit. BP
bob pruzek said
at 1:18 am on Feb 7, 2011
BTW, to simulate data it is probably best to use rnorm or rt functions, and not runif(n, min=0, max=1):
Real data rarely conform to what runif generates! bp
Dima Kassab said
at 1:36 am on Feb 7, 2011
Indeed, these questions are topics any one of our small group (Catherine, Dima and Tony) encounters and doesn't have a clear idea about! We post them here to keep track about areas we need to learn more about. Eventually if no one of our group has an answer and the questions aren't covered in class we'll make the q's explicit.
Thanks for your concern Bob.
You don't have permission to comment on this page.