ver. 1.0.0

Previous wrong code // ss_doubly_indexed[cd,jjj]=(hits_post_ext_doubly_indexed[C+1-cd,jjj]-NL*p_rev_Extented[cd])^2/(NL*p_rev_Extented[cd]); Current correct code ss_doubly_indexed[cd,jjj]=(hits_post_ext_doubly_indexed[cd,jjj]-NL*p_rev_Extented[cd])^2/(NL*p_rev_Extented[cd]);//2022Jan20

Previous wrong code // xx[cd]=(h[C+1-cd]-NL*p_rev_Extented[cd])^2/(NL*p_rev_Extented[cd]); Current correct code xx[cd]=(h[cd]-NL*p_rev_Extented[cd])^2/(NL*p_rev_Extented[cd]);//2022Jan20

This correction is very small but important, so this is the reason of the “major” update.

As one of the patients with chemical sensitivity, I strongly criticize companies that make volatile organic compounds such as synthetic detergents, pesticides, and deodorants.

But, Stan is not so good to calculate p value. Because even if R does not launch NaN in some caculation, stan launched NaN and which causes that fitted model object seems not reliable. Even if MCMC sample of model parameter is drawn well, there is a case that generative quantities block does not get good samples (including NaN). In this case, I calculate by R with samples of model parameter then the calculation is good even if calculation of generative quantities block is bad.

Prof. R. Lang does not “run as administrator” but he try to install packages in C:\Program Files\R\R-4.1.2\library which allows for administrator.

So the following error occurs is not writable Error in install.packages("devtools") : unable to install packages

Moreover, in case of windows, when the above adiministrator issues occurs, Prof. R lang suggests to make a directory to install pkgs on Onedrive, which also fails for non English languages. So, in default, R fails to install pkgs and Prof. R Lang does not know but he let us to struggle for this issues. I hope the directory suggtested by R is not on Onedrive. When I install pkgs on Linux (Ubuntu distribution) on vbox, of course Onedrive issue does not occur.

One solution: Set variable R_LIBS_USER and path C:/Users/tsuno/aaa in a writable directory (e.g., C:/Users/tsuno/lib) for non administrators. Such a directory is not automatically made, so I have to make it. Moreover, if I declare .libPaths("C:/Users/tsuno/bbb" ) in the .Rprofile in project, then .Rprofile and the above path should be same, i.e., aaa = bbb. In default, R try to install pkgs into Onedrive but it is problematic. Onedrive is not good for non-English users.

When I use devtools::release(), I have to install git (not from R console but from web page) If not then

Error in system2("git", c("rev-parse", "--abbrev-ref", "HEAD"), stdout = TRUE) : 
  '"git"' not found

Before submission, remove files inst > extdata > ****.rds

ver. 0.5.0

\[\theta_1, \theta_2,... \sim \pi(\theta|y)\] and samples from models, namely,

\[ y_{1,1},y_{1,2},y_{1,3}, ....,y_{1,N} \sim l(y|\theta_1),\] \[ y_{2,1},y_{2,2},y_{2,3}, ....,y_{2,N} \sim l(y|\theta_2),\] \[ y_{3,1},y_{3,2},y_{3,3}, ...,y_{3,N} \sim l(y|\theta_3),\] \[ y_{4,1},y_{4,2},y_{4,3}, ....,y_{4,N} \sim l(y|\theta_4),\] \[:\] \[:\]

In the previous version, I implemented in case of N = 1, only. But it was not sufficient theoretically but maybe practically sufficient? So, in the current version, we use arbitrary N in the function BayesianFROC::fit_GUI_Shiny(). This was done, 1 years ago. but I do not upload because everything seemed boring about this package.

-usethis::edit_rstudio_snippets() delete all default values, why? this is not good…. but return value gives its path, this is nice.

The following let rJava work fine, but I am not sure this is effective for others.


Why Sys.getenv() shows all paths and sometimes contains ?????? which is non-English words, for example,

  HOME                             C:/Users/A_user/OneDrive/?????? <- ????? is not English, so causes issues
  R_USER                           C:/Users/A_user/OneDrive/??????

To correct the OneDrive issues, I have to run the following code on R(Studio) console

 # work around mis-encoded environment variables
HOME <- file.path(USERPROFILE, "Documents", fsep = "\\")
R_LIBS_USER <- file.path(HOME, "R", "win-library", getRversion()[1, 1:2], fsep = "\\")
# update library paths
if (!isTRUE(file.info(R_LIBS_USER)$isdir))
  dir.create(R_LIBS_USER, recursive = TRUE)

message(Sys.getenv(“HOME”)) message(Sys.getenv(“R_USER”))

message(Sys.getenv(c(“HOME”, “R_USER”)))

I had forgetten the following code before submission, Oh my gosh.

Before submission, what I have to do are the followings

Future development plan of the author’s life

As a detoxification in our body, there is the notion of conjugation. To help this, vitamin and minerals are recommended. Manufactures never accept the toxicity of their products such as syndet.

Future development plan of this package

Reference Bayesian analysis of a ROC curve for categorical data using a skew-binormal model; Balgobin Nandram and Thelge Buddika Peiris (This paper is nice!); 2018 volume 11 369-384; Statistics and its interface

Explanatory Variables

\[d \log \Phi \in \text{Exponential family}?\] If not how we can approximate it by exponential family

If x1(t),…,xn(t) are curves into plane, then we can define its mean curve

by x(t):= mean(x1,…,xn). Or use mean parameter over all readers to get modality-wise one.

I think, if number of readers are very large, then I guess such pooling will be required. 2020 Jan

ver. 0.4.0

Major revision

Using generated quantities block in Stan files, calculation of posterior predictive p values are dratistically improved. The result is reasonable. Also, in MRMC, the author wrote down the model by using equivalent distributions, namely, multi-nomial parts.

Minor comments

x[1] <- x[3]

x[2] <- x[2]

x[3] <- x[1] (<- x[3])

therefore, the resulting vector is x[3],x[2],x[3].

I suffer MCS, this is the very bad world line. I hope that I live in another world sheet. I guess my dirchlet initial value is not so good. Bad D brane.

In roxygen comment, \code{} should not be in mutiple lines but the author always do this and R is nagging. Sorry, Prof. R Lang.

ver. 0.3.1

ver. 0.3.0

I diseased from multiple chemical sensitivity which was very heavy, and, now, my body still a lot of prurigo nodularises, chronic inflammations. The initial toxicant is synthetic detergent (i.e., syndet). When I made this pkg, I hope programming and statistics save my life, but it does not. Above me only sky.

ver. 0.2.3

Check Package –R CMD check additional options should be specified as following

   --as-cran --run-donttest

ver. 0.2.2

ver. 0.2.1

ver. 0.2.0

Moreover such multiple line cannot be detected by the R CMD check in my computer but in R CMD check in CRAN detects it. So the debug or find such multiple line is very hard to find because the error message never specify the information about such a location.

ver. 0.1.5

(Now 2020 AUgust, these are replaced by fit_GUI_Shiny() instead )

ver. 0.1.4

which cause the error

           [1] "Error in sampler$call_sampler(args_list[[i]]) : Initialization failed."

Some Stan developer taught this in stack over flows, His answer is very plain, I appreciate him. He helps me many times, I appreciate him.

ver. 0.1.3

ver. 0.1.2

ver. 0.1.1