A Note on Frequency Distributions:

Obviously a histogram will divide your data into bins. Here I have an imaginary data called “dataname” (actually the kfm$weight from lab)

• As shown in the RLab (www.stanford.edu/~epurdom/RLab.htm) you can change the number of bins by specifying the breaks

>hist(dataname, breaks=10)

• If I want to retrieve the information used in making the histogram, like the frequencies, the midpoints, etc. (like for the homework), I instead want to assign the output of the histogram to a variable name to save.

>histinfo<-hist(dataname, breaks=10)

If I type “histinfo” at the prompt, I will see all of the information that this object I created contains:

>histinfo

> histinfo

$breaks

[1] 3.999999 4.200001 4.400001 4.600001 4.800001 5.000001 5.200001 5.400001

[9] 5.600001 5.800001 6.000001 6.200001 6.400001 6.600001

$counts

[1] 1 3 0 6 3 8 8 6 6 3 4 0 2

$intensities

[1] 0.09999934 0.30000000 0.00000000 0.60000000 0.30000000 0.80000000

[7] 0.80000000 0.60000000 0.60000000 0.30000000 0.40000000 0.00000000

[13] 0.20000000

$density

[1] 0.09999934 0.30000000 0.00000000 0.60000000 0.30000000 0.80000000

[7] 0.80000000 0.60000000 0.60000000 0.30000000 0.40000000 0.00000000

[13] 0.20000000

$mids

[1] 4.100000 4.300001 4.500001 4.700001 4.900001 5.100001 5.300001 5.500001

[9] 5.700001 5.900001 6.100001 6.300001 6.500001

$xname

[1] "kfm$weight"

$equidist

[1] TRUE

attr(,"class")

[1] "histogram"

• I want to use this information, say to find the mean of the freq. distribution (i.e. pretend that the data was actually equal to the midpoints of the bin they were assigned to). So I can call the midpoints and the counts to get a weighted average (multiplying the midpoints times their frequencies)

>sum(histinfo$counts*histinfo$mids)/50

I could similarly do this calculation to get a weighted SD and V