[R] table data as input to the function

[Jim Lemon]

On 02/24/2010 04:30 PM, chinna wrote:
>
> sorry for asking again and again
>
> my Requirement:
>
> i am connecting to teradata database and i am accessing tables and table
> data also, i need generate graphs using that data and also i need  to
> forecast the results.
>
> for example
>
> i have a table xyz
>
> Store      Year       Revenue
>
> abc         2010       $557889
> def          2010       $697356
>
> i want to draw a barplot and i want to forecast the results  of revenue for
> next year
>
>
> can u please tell me wheather this is possible or not?
> if possible give me the suggestions.
>
>
> (graphs in the sense using some business intelligence tools like
> cognos,business objects using that tools we are generating graphs for
> analysing  sales information of retail stores and graphs generation.)
>

Hi chinna,
Drawing a barplot is relatively easy, as there are quite a few functions 
apart from plain old "barplot" that will give you a lot of options. Here 
is a candy-colored example using barNest in the plotrix package:

xyz<-data.frame(Store=rep(c("abc","def","ghi"),3),
  Year=rep(2007:2009,each=3),Revenue=rnorm(9,500000,100000))
barNest(Revenue~Year+Store,xyz,showall=TRUE,
  main="Revenue by store and year",
  ylab="Revenue",col=list("gray",2:4,5:7))

The forecasting part could be as simple as:

abc.lm<-lm(Revenue~Year,xyz[xyz$Store == "abc",])
def.lm...

and then use the coefficient for Year to provide a linear estimate of 
the succeeding year(s). For the toy data I generated, I estimated that 
store abc would make a loss of $94554 in 2010.


If you want more than a simple linear predictor, just use a different 
model.

Jim