From: Benjamin Auder Date: Wed, 22 Feb 2017 09:32:16 +0000 (+0100) Subject: all set-up to prepare ultimate test before last reports X-Git-Url: https://git.auder.net/%7B%7B%20asset%28%27mixstore/css/img/doc/current/gitweb.css?a=commitdiff_plain;h=613a986f1517ab5a641771da62eba6c7f5a41577;p=talweg.git all set-up to prepare ultimate test before last reports --- diff --git a/pkg/data/README b/data/README similarity index 100% rename from pkg/data/README rename to data/README diff --git a/data/data.tar.xz b/data/data.tar.xz new file mode 100644 index 0000000..785b39f --- /dev/null +++ b/data/data.tar.xz @@ -0,0 +1 @@ +#$# git-fat 9d5a273f6f71606d55f37372d1eaf0d8a7fee3f6 609260 diff --git a/pkg/data/scripts/augment_meteo.R b/data/scripts/augment_meteo.R similarity index 100% rename from pkg/data/scripts/augment_meteo.R rename to data/scripts/augment_meteo.R diff --git a/data/scripts/build_testdata.R b/data/scripts/build_testdata.R new file mode 100644 index 0000000..9bc4fa4 --- /dev/null +++ b/data/scripts/build_testdata.R @@ -0,0 +1,6 @@ +# Only one covariable (so that the "matrix" is always invertible) + +# Series are 1,1,...,1 2,2,...,2 ... 7,7,...,7 1,1,...,1 ...etc +# First serie on monday 2007-01-01 + +#TODO: diff --git a/pkg/data/scripts/fill_NAs.R b/data/scripts/fill_NAs.R similarity index 100% rename from pkg/data/scripts/fill_NAs.R rename to data/scripts/fill_NAs.R diff --git a/pkg/DESCRIPTION b/pkg/DESCRIPTION index 2a29f90..1453d54 100644 --- a/pkg/DESCRIPTION +++ b/pkg/DESCRIPTION @@ -15,7 +15,8 @@ Suggests: roxygen2, testthat, rmarkdown, - rainbow + rainbow, + knitr LazyData: yes URL: http://git.auder.net/?p=talweg.git License: MIT + file LICENSE @@ -36,3 +37,4 @@ Collate: 'getForecast.R' 'plot.R' 'utils.R' +VignetteBuilder: knitr diff --git a/pkg/R/F_Neighbors.R b/pkg/R/F_Neighbors.R index 7144fad..3abd0ff 100644 --- a/pkg/R/F_Neighbors.R +++ b/pkg/R/F_Neighbors.R @@ -173,7 +173,7 @@ NeighborsForecaster = setRefClass( } sigma = cov(M) #NOTE: robust covariance is way too slow - sigma_inv = qr.solve(sigma) + sigma_inv = solve(sigma) #TODO: use pseudo-inverse if needed? # Distances from last observed day to days in the past distances2 = rep(NA, nrow(M)-1) diff --git a/pkg/R/getData.R b/pkg/R/getData.R index 205ee5d..574522b 100644 --- a/pkg/R/getData.R +++ b/pkg/R/getData.R @@ -17,6 +17,11 @@ #' #' @return An object of class Data #' +#' @example +#' ts_data = read.csv(system.file("extdata",ts_data,package="talweg",mustWork=TRUE)) +#' exo_data = read.csv(system.file("extdata",exo_data,package="talweg",mustWork=TRUE)) +#' getData(ts_data, exo_data, ...) +#' #' @export getData = function(ts_data, exo_data, input_tz="GMT", date_format="%d/%m/%Y %H:%M", working_tz="GMT", predict_at=0, limit=Inf) @@ -28,7 +33,7 @@ getData = function(ts_data, exo_data, input_tz="GMT", date_format="%d/%m/%Y %H:% working_tz = working_tz[1] if ( (!is.data.frame(ts_data) && !is.character(ts_data)) || (!is.data.frame(exo_data) && !is.character(exo_data)) ) - stop("Bad time-series / exogenous input (data [frame] or CSV file)") + stop("Bad time-series / exogenous input (data frame or CSV file)") if (is.character(ts_data)) ts_data = ts_data[1] if (is.character(exo_data)) @@ -41,29 +46,15 @@ getData = function(ts_data, exo_data, input_tz="GMT", date_format="%d/%m/%Y %H:% date_format = date_format[1] ts_df = - if (is.character(ts_data)) { - if (ts_data %in% data(package="talweg")$results[,"Item"]) - ts_data = - - - - - ############CONTINUE: http://r-pkgs.had.co.nz/data.html - - - - - - read.csv(ts_data) - } else { + if (is.character(ts_data)) + read.csv(ts_data) + else ts_data - } exo_df = - if (is.character(exo_data)) { + if (is.character(exo_data)) read.csv(exo_data) - } else { + else exo_data - } # Convert to the desired timezone (usually "GMT" or "Europe/Paris") formatted_dates_POSIXlt = strptime(as.character(ts_df[,1]), date_format, tz=input_tz) ts_df[,1] = format(as.POSIXct(formatted_dates_POSIXlt), tz=working_tz, usetz=TRUE) diff --git a/pkg/R/getForecast.R b/pkg/R/getForecast.R index b48f3da..6e7720e 100644 --- a/pkg/R/getForecast.R +++ b/pkg/R/getForecast.R @@ -24,7 +24,7 @@ #' @return An object of class Forecast #' #' @examples -#' data = getData(ts_data="data/pm10_mesures_H_loc.csv", exo_data="data/meteo_extra_noNAs.csv", +#' data = getData(ts_data="pm10_mesures_H_loc.csv", exo_data="meteo_extra_noNAs.csv", #' input_tz = "Europe/Paris", working_tz="Europe/Paris", predict_at=7) #' pred = getForecast(data, 2200:2230, "Persistence", "Persistence", 500, 12) #' \dontrun{#Sketch for real-time mode: diff --git a/pkg/data/data.tar.xz b/pkg/data/data.tar.xz deleted file mode 100644 index 6e5e696..0000000 --- a/pkg/data/data.tar.xz +++ /dev/null @@ -1 +0,0 @@ -#$# git-fat fb2d21849524e743ce1b3e5589efb40d1182f468 564468 diff --git a/pkg/inst/CITATION b/pkg/inst/CITATION new file mode 100644 index 0000000..2167e9e --- /dev/null +++ b/pkg/inst/CITATION @@ -0,0 +1,17 @@ +citHeader("To cite talweg in publications use:") + +citEntry(entry = "Manual", + title = "Time-series sAmpLes forecasted With ExoGenous variables.", + author = personList(as.person("Benjamin Auder"), + as.person("Jean-Michel Poggi"), + as.person("Bruno Portier")), + organization = "Paris-Sud & INSA Rouen", + address = "Orsay & Rouen, France", + year = "2017", + url = "https://git.auder.net/?p=talweg.git", + + textVersion = + paste("Benjamin Auder, Jean-Michel Poggi, Bruno Portier (2017).", + "talweg: Time-series sAmpLes forecasted With ExoGenous variables.", + "URL https://git.auder.net/?p=talweg.git") +) diff --git a/pkg/data/meteo_extra_noNAs.csv b/pkg/inst/extdata/meteo_extra_noNAs.csv similarity index 100% rename from pkg/data/meteo_extra_noNAs.csv rename to pkg/inst/extdata/meteo_extra_noNAs.csv diff --git a/pkg/data/pm10_mesures_H_loc.csv b/pkg/inst/extdata/pm10_mesures_H_loc.csv similarity index 100% rename from pkg/data/pm10_mesures_H_loc.csv rename to pkg/inst/extdata/pm10_mesures_H_loc.csv diff --git a/pkg/inst/testdata/exo.csv b/pkg/inst/testdata/exo.csv new file mode 100644 index 0000000..e69de29 diff --git a/pkg/inst/testdata/ts.csv b/pkg/inst/testdata/ts.csv new file mode 100644 index 0000000..e69de29 diff --git a/pkg/tests/testthat/test.Forecaster.R b/pkg/tests/testthat/test.Forecaster.R index b20f104..0986a3d 100644 --- a/pkg/tests/testthat/test.Forecaster.R +++ b/pkg/tests/testthat/test.Forecaster.R @@ -2,34 +2,34 @@ context("Check that forecasters behave as expected") test_that("Average+Zero method behave as expected", { + ts_data = system.file("testdata","ts",package="talweg") + exo_data = system.file("testdata","exo",package="talweg") + + data0 = getData(ts_data, exo_data, input_tz="GMT", date_format="%Y-%m-%d %H:%M", + working_tz="GMT", predict_at=0, limit=Inf) + #TODO: with and without shift at origin (so series values at least forst ones are required) + + indices = ... + pred0 = getForecast(.......) + + for (i in seq_along(indices)) + { + expect_identical(....) + } + + data13 = getData(ts_data, exo_data, input_tz="GMT", date_format="%Y-%m-%d %H:%M", + working_tz="GMT", predict_at=13, limit=Inf) + #Attention: jours deviennent 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 (14h-->0h puis 1h-->13h) +} test_that("Persistence+Zero method behave as expected", { +} test_that("Neighbors+Zero method behave as expected", { +} test_that("Neighbors+Neighbors method behave as expected", { - - - -#TODO: with and without shift at origin (so series values at least forst ones are required) - - - n = 1500 - series = list() - for (i in seq_len(n)) - { - index = (i%%3) + 1 - level = mean(s[[index]]) - serie = s[[index]] - level + rnorm(L,sd=0.05) - # 10 series with NAs for index 2 - if (index == 2 && i >= 60 && i<= 90) - serie[sample(seq_len(L),1)] = NA - series[[i]] = list("level"=level,"serie"=serie) #no need for more :: si : time !!! - } - data = new("Data", data=series) - - dateIndexToInteger = function(index, data) -}) +} diff --git a/pkg/tests/testthat/test.dateIndexToInteger.R b/pkg/tests/testthat/test.dateIndexToInteger.R index 7f09847..b2316dd 100644 --- a/pkg/tests/testthat/test.dateIndexToInteger.R +++ b/pkg/tests/testthat/test.dateIndexToInteger.R @@ -1,33 +1,25 @@ context("Check that dateIndexToInteger behaves as expected") -getDataTest = function(n, shift) +test_that("computed integer index is correct; predict_at == 0", { - n = 1500 - series = list() - s = rep(0, 24) - - for (i in seq_len(n)) - { - level = i %% 3mean(s[[index]]) - serie = s - # 10 series with NAs for index 2 - if (index == 2 && i >= 60 && i<= 90) - serie[sample(seq_len(L),1)] = NA - series[[i]] = list("level"=i%%3, "serie"=s, "time"=) - } - data = new("Data", data=series) -} - -test_that("integer index matches date in data, predict_at == 0", -{ - data = getData( - dateIndexToInteger = function(index, data) + data0 = getData(ts_data="pm10_mesures_H_loc.csv", exo_data="meteo_extra_noNAs.csv", + input_tz="Europe/Paris",working_tz="Europe/Paris", predict_at=0, limit=200) + expect_identical( dateIndexToInteger("2008-12-10",data), 1 ) + expect_identical( dateIndexToInteger("2008-12-11",data), 2 ) + expect_identical( dateIndexToInteger("2008-12-20",data), 11 ) + expect_identical( dateIndexToInteger("2009-02-01",data), 53 ) + expect_identical( dateIndexToInteger("2009-03-01",data), 81 ) + expect_identical( dateIndexToInteger("2009-05-31",data), 172 ) }) -test_that("integer index matches date in data, predict_at > 0", +test_that("computed integer index is correct; predict_at > 0", { - - - - - ####TODO: CSV as raw data in inst/extdata http://r-pkgs.had.co.nz/data.html + data7 = getData(ts_data="pm10_mesures_H_loc.csv", exo_data="meteo_extra_noNAs.csv", + input_tz="Europe/Paris",working_tz="Europe/Paris", predict_at=7, limit=200) + expect_identical( dateIndexToInteger("2008-12-10",data), 2 ) + expect_identical( dateIndexToInteger("2008-12-11",data), 3 ) + expect_identical( dateIndexToInteger("2008-12-20",data), 12 ) + expect_identical( dateIndexToInteger("2009-02-01",data), 54 ) + expect_identical( dateIndexToInteger("2009-03-01",data), 82 ) + expect_identical( dateIndexToInteger("2009-05-31",data), 173 ) +} diff --git a/pkg/tests/testthat/test.integerIndexToDate.R b/pkg/tests/testthat/test.integerIndexToDate.R index 31cd740..5f3d007 100644 --- a/pkg/tests/testthat/test.integerIndexToDate.R +++ b/pkg/tests/testthat/test.integerIndexToDate.R @@ -1,22 +1,25 @@ context("Check that integerIndexToDate behaves as expected") -test_that("date matches index in data", +test_that("computed dates match the input indexes; predict_at == 0", { - #TODO: with and without shift at origin (so series values at least forst ones are required) - - n = 1500 - series = list() - for (i in seq_len(n)) - { - index = (i%%3) + 1 - level = mean(s[[index]]) - serie = s[[index]] - level + rnorm(L,sd=0.05) - # 10 series with NAs for index 2 - if (index == 2 && i >= 60 && i<= 90) - serie[sample(seq_len(L),1)] = NA - series[[i]] = list("level"=level,"serie"=serie) #no need for more :: si : time !!! - } - data = new("Data", data=series) - - integerIndexToDate = function(index, data) + data0 = getData(ts_data="pm10_mesures_H_loc.csv", exo_data="meteo_extra_noNAs.csv", + input_tz="Europe/Paris",working_tz="Europe/Paris", predict_at=0, limit=200) + expect_identical( integerIndexToDate(1,data), as.Date("2008-12-10") ) + expect_identical( integerIndexToDate(2,data), as.Date("2008-12-11") ) + expect_identical( integerIndexToDate(11,data), as.Date("2008-12-20") ) + expect_identical( integerIndexToDate(53,data), as.Date("2009-02-01") ) + expect_identical( integerIndexToDate(81,data), as.Date("2009-03-01") ) + expect_identical( integerIndexToDate(172,data), as.Date("2009-05-31") ) }) + +test_that("computed dates match the input indexes; predict_at > 0", +{ + data7 = getData(ts_data="pm10_mesures_H_loc.csv", exo_data="meteo_extra_noNAs.csv", + input_tz="Europe/Paris",working_tz="Europe/Paris", predict_at=7, limit=200) + expect_identical( integerIndexToDate(2,data), as.Date("2008-12-10") ) + expect_identical( integerIndexToDate(3,data), as.Date("2008-12-11") ) + expect_identical( integerIndexToDate(12,data), as.Date("2008-12-20") ) + expect_identical( integerIndexToDate(54,data), as.Date("2009-02-01") ) + expect_identical( integerIndexToDate(82,data), as.Date("2009-03-01") ) + expect_identical( integerIndexToDate(173,data), as.Date("2009-05-31") ) +} diff --git a/pkg/vignettes/talweg.Rmd b/pkg/vignettes/talweg.Rmd new file mode 100644 index 0000000..aace6af --- /dev/null +++ b/pkg/vignettes/talweg.Rmd @@ -0,0 +1,58 @@ +--- +title: "Vignette Title" +author: "Vignette Author" +date: "`r Sys.Date()`" +output: rmarkdown::html_vignette +vignette: > + %\VignetteIndexEntry{Vignette Title} + %\VignetteEngine{knitr::rmarkdown} + %\VignetteEncoding{UTF-8} +--- + +Vignettes are long form documentation commonly included in packages. Because they are part of the distribution of the package, they need to be as compact as possible. The `html_vignette` output type provides a custom style sheet (and tweaks some options) to ensure that the resulting html is as small as possible. The `html_vignette` format: + +- Never uses retina figures +- Has a smaller default figure size +- Uses a custom CSS stylesheet instead of the default Twitter Bootstrap style + +## Vignette Info + +Note the various macros within the `vignette` section of the metadata block above. These are required in order to instruct R how to build the vignette. Note that you should change the `title` field and the `\VignetteIndexEntry` to match the title of your vignette. + +## Styles + +The `html_vignette` template includes a basic CSS theme. To override this theme you can specify your own CSS in the document metadata as follows: + + output: + rmarkdown::html_vignette: + css: mystyles.css + +## Figures + +The figure sizes have been customised so that you can easily put two images side-by-side. + +```{r, fig.show='hold'} +plot(1:10) +plot(10:1) +``` + +You can enable figure captions by `fig_caption: yes` in YAML: + + output: + rmarkdown::html_vignette: + fig_caption: yes + +Then you can use the chunk option `fig.cap = "Your figure caption."` in **knitr**. + +## More Examples + +You can write math expressions, e.g. $Y = X\beta + \epsilon$, footnotes^[A footnote here.], and tables, e.g. using `knitr::kable()`. + +```{r, echo=FALSE, results='asis'} +knitr::kable(head(mtcars, 10)) +``` + +Also a quote using `>`: + +> "He who gives up [code] safety for [code] speed deserves neither." +([via](https://twitter.com/hadleywickham/status/504368538874703872))