[R] saving and reading csv and rda files

[Alok Bohara, PhD]

Hi Michael:

Thanks for your patience in dealing with my questions.  Now, I have a 
slightly better sense of R environment...  data frame versus object 
versus names versus tail and head  etc.. etc...

I have been an avid user of  Gauss  (and Stata).  And, I have recently 
begun to use  R for my research.  I had used R for various optimization 
problems using the straight csv data files/frames, and I never paid any 
attention to object etc..     I used Stata to do my data preparation... 
   In any case, thanks for your patience in your explanation...

Best,
Alok


On 8/27/2012 2:04 PM, R. Michael Weylandt wrote:
> On Mon, Aug 27, 2012 at 9:56 AM, Alok Bohara, PhD <bohara at unm.edu> wrote:
>> Hi:
>>
>> I am trying to understand the link between ".csv" and  ".rda" files.    Is
>> there any easy to follow tutorial on this?
>>
>> (I could do some of the operations below, but I got lost in the details.)
>>
>> 1.  Reading .rda file ?
>>
>> data <- load("profit.rda")         # supposed to have four variable --y x1
>> x2 state_name
>>
>> --how do I find out about the variable names,
>> --take the log of y and x1
>> --extract y and calculate mean etc..
>>
>> (I want to use it in lm regression lny = f(lnx1,x2))
> I personally find load() a little dangerous. It reloads objects with
> the names they had when you saved them, potentially clobbering other
> objects by that name. I'd recommend saveRDS/readRDS personally since
> they don't clobber (they instead do what I think you think load()
> does). That said,
>
> data <- load(...) actually doesn't put the variables in data; rather
> it puts them into the global envir by their original name and puts the
> names as strings into data. I'd guess if you type data, you'll find
> something like
>
>> data
> [1] "y" "x1" "x2", "state_name"
>
> You probably don't want to use "data" as a name in real code, because
> it's the name of an important function used to, not surprisingly, load
> data.
>
> As far as the model fitting code, that should be covered in most basic
> R tutorials, but you're looking for something like
>
> lm( log(y) ~ x1)
>
> Finally, I think you're a little tripped up on the multiple uses of
> the term "names". "name" most commonly refers to variable name of an
> object, but you seem to be confusing it with column names, which are
> often recording what the variables in a data set are. The data set as
> a unit has a "name" and then each of the recorded variates corresponds
> to a column name. when you read/write csv files with header = TRUE,
> the text keeps the column names, not the R object names.
>
>
>> ****************
>>
>> 2. How could I save this profit.rda file as a csv file  with the variable
>> names attached?
>> I tried doing this:
>>
>> profit_data <- load("profit.rda")
> Again, this doesn't work because load() doesn't return the loaded
> objects, only the name.
>
>> #Could I do this?
>> write.csv(profit_data, file="profit.csv", col.names=TRUE)
>> data2 <- read.csv("profit.csv", head = TRUE)
> This will work in that profit_data == data2, but profit_data isn't the
> desired result of load("profit.rda").
>
>> # saving .rda file without the header?
>> write.csv(profit_data, file="profit2.csv", col.names=FALSE)
>>
>
>
>> ***********
>>
>> 3. Creating a .rda file  from a csv file  using the save command
>>
>> data2 <- read.csv("poverty.csv", head = T, sep = ",")   # poverty.csv file
>> has 4 variables -- Q L K  country_name
>> save(data2, file="poverty.rda")
>>
>> --How do I  attach names from the csv file to this .rda file?
> Take a look at data2: it probably has "Q","L","K","country_name" as
> _column_ names (not the same as object names). If it does, the
> save/load cycle will preserve them.
>
> Cheers,
> Michael
>
>>
>>
>> Best,
>> Alok Bohara
>> UNM
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.


-- 
Alok K. Bohara, Ph.D.
Professor
Department of Economics
MSC 05 3060
1 University of New Mexico
Albuquerque, NM 87131-0001, USA
Ph: 505-277-5903/5304(w)
Fax:505-277-9445
email: bohara at unm.edu
http://www.unm.edu/~econ/faculty/professors.html
Nepal Study Center: http://nepalstudycenter.unm.edu