Skip to content

as.Date from data.frame

5 messages · separent, Steve Lianoglou, Marc Schwartz +1 more

Original
separent
# 
#Hello,

#I loaded data using read.table - I needed to convert a row in the data
frame to date class:
V1          V2          V3          V4
1  2008-05-19  2008-04-19  2008-03-21  2008-02-22
2 38.16999817 30.70999908 36.86000061 35.18999863
3 37.47999954 29.95000076 36.45999908 35.36000061
4 37.47999954 30.14999962 36.47000122 35.36000061
5 37.84999847 30.56999969 36.84000015 35.74000168
6 38.38999939 31.14999962 37.34000015 36.27000046
7 39.11000061 31.90999985 38.02999878 36.97999954
8 39.81000137 32.65000153 38.68000031 37.63999939
9 40.47000122 33.34999847 39.27999878 38.27000046
V1         V2         V3         V4
1 2008-05-19 2008-04-19 2008-03-21 2008-02-22
do not know how to convert 'dates[1,]' to class "Date"

# However, I can individually convert them all:
[1] "2008-05-19"

# How can I change the class of the date over the entire row (my original
file contains more than one hundred rows)?
# Thank you very much.
Steve Lianoglou
# 
Hi,
On Nov 11, 2009, at 1:06 PM, separent wrote:

            

      
      
      
    

        
I think this should work if data[1,] returns a vector of characters:

R> d <- c("2008-05-19", "2008-05-30")
R> dd <- as.Date(d)
R> is(dd)
[1] "Date"     "oldClass"

Is your first row being treated as factors? Maybe if you convert to  
character first it should work?

R> dates1<-as.Date(as.character(data[1,]))

But the fact that it works for as.Date(data[1,1]) make it a bit  
weird ...

-steve

            

      
      
      
    

        

      
      
    

        
--
Steve Lianoglou
Graduate Student: Computational Systems Biology
   |  Memorial Sloan-Kettering Cancer Center
   |  Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact

  
  


  


      

    

    

      
      

        


  

        

      Marc Schwartz
    

    

      
      #
    

  


  

      
On Nov 11, 2009, at 12:06 PM, separent wrote:

        

            

      
      
      
    

        
In a data frame, you cannot change the data type for a row, without  
changing the data type for the entire column.

That being said, it looks to me like each column contains continuous  
data for a chronological series.

If that is correct, then you want the dates to be the column names and  
not the first row., Thus, when you used read.table(), you should have  
included the argument 'header = TRUE', which would do just that. With  
read.table(), header is FALSE by default.

HTH,

Marc Schwartz

  
  


  


      

    

    

      
      

        


  

        

      David Winsemius
    

    

      
      #
    

  


  

      
On Nov 11, 2009, at 1:06 PM, separent wrote:

        

            

      
      
      
    

        
You _should_ have given the actual code (as the Posting Guide  
requests), but I would guess that you should have added either  
as.is=TRUE or stringsAsFactors=FALSE. As it is, you probably have all  
factors.

            

      
      
      
    

        
You should have given the actual error message which was probably more  
informative than what your brain interpreted it to mean.

            

      
      
      
    

        
Off hand I would say you want to work on a transpose of this data. You  
cannot mix datatypes within columns and it seems fairly clear that the  
rows below number 1 are numeric. Back to the drawing board.

            

      
      
      
    

        
--

David Winsemius, MD
Heritage Laboratories
West Hartford, CT

  
  


  


      

    

    

      
      

        


  

        

      separent
    

    

      
      #
    

  


  

      
Hello Gentlemen,

All of your answers were helpfull. Indeed, 'as.is=TRUE' or
'stringsAsFactors=FALSE' avoid importing as factors, as it did when I
imported the table without specifying anything. However, data frames will
not allow different datatypes within a single column, so, for time series, I
should work with a transpose of the table. However, when the time series is
the label of a profile, I should read the table with 'header = TRUE'.

In other words:
* Plotting values of a column (let's say againts observation number) for a
selected time (where time is the name of the series): import data with
'header = TRUE'
* Plotting the evolution of the values for a single observation number
against time: 'header = TRUE' & transpose & 'as.is=TRUE'

Kind Regards,

Serge-?tienne Parent
Golder Associ?s
Canada

        
separent wrote: