ellisp / ggseas Goto Github PK

View Code? Open in Web Editor NEW

74.0 74.0 6.0 1.3 MB

seasonal adjustment on the fly extension for ggplot2

License: GNU General Public License v3.0

R 100.00%

ggseas's People

Contributors

Stargazers

Watchers

Forkers

paulhendricks fsgp rlugojr pitakakariki y461650833y wangrenfeng0

ggseas's Issues

new ggplot2 breaks ggsdc (and maybe other parts of the package)

see https://www.tidyverse.org/articles/2018/05/ggplot2-2-3-0/

calculate frequency and start from posixct dates

Needs checking and testing but I think currently this isn't done

move packages from 'depends' to 'imports'

This might or might not help with #15 but it would be good practice, instead of importing all of seasonal and stats, to import just the functions needed.

for some reason the CRAN version isn't working, at least on one of my machines. Returns errors even when running the basic examples of stat_stl, stat_decomp, etc. Hopefully this is temporary. In the meantime, the github version (which is identical in theory) works fine.

Is there any way to change facet names in ggsdc?

My native language in Portuguese and I'm writing my undergraduate thesis in which I will show a plot made by using ggsdc(). I wonder if I can manually change the facets names so I can write what they mean in Portuguese.

add option to convert to an index

Faceted decomposition graphs for Classic, STL, and X-13 ARIMA SEATS similar to default plotting of class "stl"

Great package. What would you think of possibly implementing functionality that provides a facet of decomposition similar to the default plotting of an "stl" class.

In code, something that mimics this:

some_ts <- ts(rnorm(100), frequency = 12)
plot(stl(some_ts, s.window = "periodic"))

But with ggplot2 styles and similar to your API. Maybe a facet = TRUE flag?

library(ggseas)
# make demo data
ap_df <- data.frame(
   x = as.numeric(time(AirPassengers)),
   y = as.numeric(AirPassengers)
)

ggplot(ap_df, aes(x = x, y = y)) +
   geom_line(colour = "grey80") +
   stat_decomp(frequency = 12, facet = TRUE)

New ggplot broke ggseas?

I'm getting this error in the first time using ggseas after upgrading to R 3.3.2 and ggplot2 2.2.0

Error: StatSeas was built with an incompatible version of ggproto.
Please reinstall the package that provides this extension.
Execution halted

Exited with status 1.

Add support for rolling averages

stat_rollapplyr

Warning on OSX and Solaris with vignette build

https://cran.r-project.org/web/checks/check_results_ggseas.html

Version: 0.5.1 
Check: re-building of vignette outputs 
Result: WARN 
    Error in re-building vignettes:
     ...
    Warning in engine$weave(file, quiet = quiet, encoding = enc) :
     Pandoc (>= 1.12.3) and/or pandoc-citeproc not available. Falling back to R Markdown v1.
    Quitting from lines 125-126 (ggsdc.Rmd) 
    Error: processing vignette 'ggsdc.Rmd' failed with diagnostics:
    Process terminated
    Execution halted 
Flavor: r-patched-solaris-x86

Version: 0.5.1 
Check: re-building of vignette outputs 
Result: WARN 
    Error in re-building vignettes:
     ...
    
    Backtrace for this error:
    #0 0x129b9e579
    #1 0x129b9d945
    #2 0x7fff93ac4529
    sh: line 1: 89501 Trace/BPT trap: 5 /Volumes/SSD-Data/Builds/R-dev-web/QA/Simon/packages/el-capitan-x86_64/Rlib/3.5/x13binary/bin/x13ashtml /var/folders/xn/m8dzdsgs7sg8q7jgflnqm8w80000gn/T//RtmppExnyS/x1315bc558df5aa/iofile -n -s 2> /dev/null
    dyld: lazy symbol binding failed: Symbol not found: _mkostemp
     Referenced from: /Volumes/SSD-Data/Builds/R-dev-web/QA/Simon/packages/el-capitan-x86_64/Rlib/3.5/x13binary/bin/../lib/libgfortran.3.dylib (which was built for Mac OS X 10.12)
     Expected in: /usr/lib/libSystem.B.dylib
    
    dyld: Symbol not found: _mkostemp
     Referenced from: /Volumes/SSD-Data/Builds/R-dev-web/QA/Simon/packages/el-capitan-x86_64/Rlib/3.5/x13binary/bin/../lib/libgfortran.3.dylib (which was built for Mac OS X 10.12)
     Expected in: /usr/lib/libSystem.B.dylib
    
    
    Program received signal SIGTRAP: Trace/breakpoint trap.
    
    Backtrace for this error:
    #0 0x125f5f579
    #1 0x125f5e945
    #2 0x7fff93ac4529
    Quitting from lines 125-126 (ggsdc.Rmd) 
    Error: processing vignette 'ggsdc.Rmd' failed with diagnostics:
    X-13 has returned a non-zero exist status, which means that the current spec file cannot be processed for an unknown reason.
    Execution halted 
Flavor: r-release-osx-x86_64

changing the order of levels of the factor mapped to colour doesn't control it in legend

in ggsdc

     ldeaths_df %>%
        mutate(sex = factor(sex, levels = c("male", "female"))) %>%
   ggsdc(aes(x = YearMon, y = deaths, colour = sex), method = "seas") +
   geom_line()

and

     ldeaths_df %>%
        mutate(sex = factor(sex, levels = c("female", "male"))) %>%
   ggsdc(aes(x = YearMon, y = deaths, colour = sex), method = "seas") +
   geom_line()

both produce the same results and they should change the order on the legend. An "as character" somewhere is killing the factor level ordering.

ggsdc doesn't play nicely with tibble

For exmaple (not reproducible unless you have cam_df2):

This works:

cam_df2 %>%
  data.frame() %>%
  ggsdc(aes(x = TimePeriod, y = Value, color = CV1), start = c(1996,7), frequency = 12, s.window = 7) +
  geom_line()

but this doesn't:

cam_df2 %>%
  ggsdc(aes(x = TimePeriod, y = Value, color = CV1), start = c(1996,7), frequency = 12, s.window = 7) +
  geom_line()

Doesn't work unless source data are in time order

ldeaths_sorted <- ldeaths_df[order(ldeaths_df$deaths), ]

  ggplot(ldeaths_sorted, aes(x = YearMon, y = deaths, colour = sex)) +
       stat_decomp(frequency = 12)

The fix will be to sort it by x during the stat.

Problem building vignette for Mac on CRAN

https://www.r-project.org/nosvn/R.check/r-devel-macos-x86_64-clang/ggseas-00check.html

checking re-building of vignette outputs ... WARNING
Error in re-building vignettes:
  ...
Quitting from lines 125-126 (ggsdc.Rmd) 
Error: processing vignette 'ggsdc.Rmd' failed with diagnostics:
Process terminated
Execution halted

ggsdc has an example that takes >5 seconds to run on win_build

possibly good to get this down to 4 seconds :)

Solaris build on CRAN has a problem with vignette

Add ability to export to plotly

"To add support for your custom geom, say GeomCustom, y ou'll need to export a to_basic.GeomCustom in your extension package. For example, since GeomLine is a special case of GeomPath sorted by x-value, to add support for GeomLine, we just sort the data by x and prefix a GeomPath class. If you can't convert your custom geom to a "basic" geom, you might need to also export a geom2trace.GeomCustom method. This will require some knowledge of the plotly.js JSON spec, but we'd be more than happy to help. If anyone has questions, or suggestions on better ways to do things, please let me know!"

make proper testthat testing harness

some of the examples use dplyr::filter without importing it

probably can just change to subset?