petolau / tsrepr Goto Github PK

I saw that you calculate MASE by "error = diff / denom;". However, according to Hyndman, Koehler (2006) your "denom" should be defined as (1/ (N-m)) * sum(abs( korrekt(i) - naive(i-m))) (with m being the time-shift). I think your calculation misses this.

Hi, I was using your repr_sma function but the result was not consistent with the input, then I tried it with simple examples and still got unexpected results.

This is detailed feedback for your JOSS review openjournals/joss-reviews#577.

R packages should have a tests/ directory containing unit tests. Hadley Wickham’s R packages book has a whole chapter explaining how to do this.

Better yet, you could set up a continuous integration system with Travis CI and add a badge to your README.md that shows that your package passes all of the tests.

Dear @PetoLau

I have 298 time-series dataset, may ask you about how do that like a return dataframe. of all time-series.

b_repr_df<- reinterpolate(b_lst_df[2L:298L], new.length = max(lengths(b_lst_df[2L:298L])))
b_repr_df<- repr_pla(b_lst_df[2L:298L],time=1090)

it will show an error like this

Error in repr_pla(b_lst_df[2L:297L]) : 
  (list) object cannot be coerced to type 'double'

but if i run one by one time-series, it is no problem.
b_repr_df<- repr_pla(b_lst_df$coin, time= 1090, return='both')

when i write below line, it produces the error in loadnamespace
library("TSrepr", lib.loc="~/R/win-library/3.2")
Error in loadNamespace(j <- i[[1L]], c(lib.loc, .libPaths()), versionCheck = vI[[j]]) :
namespace ‘Rcpp’ 0.12.10 is being loaded, but >= 0.12.12 is required
In addition: Warning message:
package ‘TSrepr’ was built under R version 3.5.1
Error: package or namespace load failed for ‘TSrepr’

Help me.

This is detailed feedback for your JOSS review openjournals/joss-reviews#577.

There’s an email contact for questions and comments, but no guidelines for contribution. I would suggest adding a CONTRIBUTING.md file to the repository. I browsed through some ROpenSci packages and many of them have guidelines like this. For example:

• Tabulizer
• Rentrez
• Rorcid

This is detailed feedback for your JOSS review openjournals/joss-reviews#577.

The paper accompanying the software is unclear because it needs to be proofread for grammar and correct wording. I can give more detailed feedback if needed.

I have weekly data that starts on a Tuesday. The output after using the repr_lm function will therefore begin on Tuesday when plotting in ggplot2. Is it possible to shift the starting day to another day (i.e., I would like the plot to start on Monday). Thank you!