sangwon-hyun / flowmix Goto Github PK

(Pie is delicious.)

For instance, |maxdev| is missing.

In get_max_lambda.Rd, Value is given as "No return", but the function actually does return a non-empty list.

Suppose we read in data via

dat <- readRDS("./flowmix/paper-data/MGL1704-hourly-paper.RDS")

dat$countslist has length 308 while dat$ylist has length 296

My quick "fix" is something like...

list2env(dat, envir = globalenv())
countslist <- countslist[names(countslist) %in% names(ylist)]

Leave the bare minimum (and the functions that were used for the manuscript); delete all else.
There are currently several R files containing plotting code; consolidate.

I get errors such as could not find function "mclapply" when I run the covarem_once function from flowcy. When I first run library(parallel) this error goes away. I believe this means you are using mclapply in the package rather than parallel::mclapply. Beyond making this change, all functions from other packages should be handled similarly. Refer to this section of Hadley Wickham's R Packages book for how to handle functions that are from other R packages.

The cv_aggregate function contains the following line of code:

stopifnot(exists(c("prob_lambdas", "mean_lambdas")))

which generates an error, since exists expects a string argument, not a vector of strings.

A workaround is to replace the above line with

stopifnot(exists("prob_lambdas"))
stopifnot(exists("mean_lambdas"))

I created an example where cv_flowmix runs into problems:
https://www.dropbox.com/s/wr1zslc775rz6i5/cv_flowmix_example.R?dl=0

nsize = pmin(nrow(y)/TT * 30, nrow(y)) reports error when TT = 300, nrow(y) = 5.