Using R, little troubles can arise. Below, a "work in progress" collection of solved problems...or not yet...
- ERRONEOUSLY SAVED WORKSPACE
If you want to delete the workspace you saved, check into the Documents folder and delete both files that appear as "R Workspace" and "Rhistory". That's all!!
- FISHER'S EXACT TEST ERROR MESSAGES: "FEXACT error 6 and 7"
Have you ever got error messages performing Fisher's exact test? Well, I have got two different error messages that I solved with the same "tricks":
Error in fisher.test(.Table) : FEXACT error 7.
LDSTP is too small for this problem.
Try increasing the size of the workspace.
OR
Error in fisher.test(.Table) : FEXACT error 6.
LDKEY is too small for this problem.
Try increasing the size of the workspace.
LDKEY is too small for this problem.
Try increasing the size of the workspace.
As suggested, you can first try to increase the workspace to 2e8 or 2e9:
fisher.test(data,workspace=2e8) #change data with the name of your table
If it still doesn't work, you can go for a simulation:
fisher.test(data,simulate.p.value=TRUE,B=1e7)
where simulate.p.value is a logical for computing p-value by Monte Carlo simulation and B indicates the number of replicates (by default B=2000), that can be changed as needed.
I know, it is not exactly the same...but at least you can run the analysis.
As usual, if you have a better solution, please let me know!
fisher.test(data,workspace=2e8) #change data with the name of your table
If it still doesn't work, you can go for a simulation:
fisher.test(data,simulate.p.value=TRUE,B=1e7)
where simulate.p.value is a logical for computing p-value by Monte Carlo simulation and B indicates the number of replicates (by default B=2000), that can be changed as needed.
I know, it is not exactly the same...but at least you can run the analysis.
As usual, if you have a better solution, please let me know!
- NEW VERSION NOT WRITABLE: " '/R-X.X.X/library' is not writable"
Have you just updated the R version but can't update the library because the folder is not writable?
Don't worry! You have two different options:
1. the easiest way is to create a private library. Just say "Yes" when R will ask you (and take note of the location of the folder);
2. if you still want to write in the library folder, you can get the authorisation following the easy instructions I provided in this post.
- ANOVA ERROR MESSAGES: "Error in eigen" OR "Error in contrasts"
If you perform a mixed ANOVA, either using aov() or anova_test() (from rstatix package), you could run into these error messages:
Error in eigen(SSPE, only.values = TRUE) :
infinite or missing values in 'x'
OR
Error in `contrasts<-`(`*tmp*`, value = contr.funs[1 + isOF[nn]]) :
contrasts can be applied only to factors with 2 or more levels
In both cases, the problem is represented by missing values (NA).
If you prepare your data excluding NA for your dependent variable with filter(), you will get the first message because there will be no values corresponding to that case in that particular level of the independent variable.
If you decide to solve the problem by eliminating the row with the NA from the dataset, you will get the second error message because R will miss completely that case in that particular level of the independent variable.
To avoid this problem you should remove the rows with that specific case at any level of the independent variable, including those where the value is not missing.
Of course, this can represent a loss of data.
One easy possibility is to substitute the NA with the mean of the group at that level of the independent variable.
However, to decide how to handle missing values, I suggest going on with further readings.
