Dear Constanze,
the error message gives you a hint, where the problem might be. The
first error massage says that you have NAs in your index variables
"Unit" and "Week". You can check that with the is.na()
function. The
second error message (r.squared) tells you to specify the model, that is
on which transformation of the data the R-squared has to be computed.
There is no default. Moreover, you insert a dataframe ("pCN04")
instead
on a plm object in r.squared().
Kind regards
Nina Sch?nfelder
-----
FernUniversit?t in Hagen
Fakult?t f?r Wirtschaftswissenschaft
Lehrstuhl f?r Volkswirtschaftslehre,
insbes. Makro?konomik
58084 Hagen
On 16/02/15 22:25, fay constanze wrote:> DearGiovanni,Congratulationsfor the?truly helpful?plm package! Being new to
R, I have a problem with the plm function for financial markets timeseries data:
After having defined a large, unbalanced panel pdata.frame
(https://www.dropbox.com/s/2r9t1cu9v65gobk/Database_CN_2004.csv?dl=0)and running
a simple OLS model of two variables regressing company returns(Perf) on index
returns (PerfIn)synch <-plm (Perf ~ PerfIn , data=pCN04, na.action = na.omit,
index=c("Unit", "Week"), model="pooling") I
keepreceiving the following error:Error in model.matrix.pFormula(formula, data,
rhs =1, model = model,? : ? NA in theindividual index variableIn
addition:Warning message:In `[.data.frame`(index, as.numeric(rownames(mf)),) :?
NAsintroduced by coercion
> There aremuch more NAs in y (Perf) than in x (PerfIn) and I have tried to
get rid ofthem with na.omit. Another error I have is with the same model for the
r.squaredfunction r.squared(pCN04,model=NULL, Perf ~ PerfIn, type=
"ess")Error in tss.default(object, model = model) : ??unused argument
(model = model)In addition: Warning message:In mean.default(haty) : argument is
not numeric or logical: returning NAMany thanks in advance,your help is very
precious to me!Constanze
>
>