🇫🇷 README en français | 🇬🇧 README in english
Présentation
Le but premier du package {JDCruncheR} est de fournir un accès rapide et facile au cruncher (JWSACruncher
) depuis R. Le cruncher est un outil de mise à jour des workspaces de JDemetra+ sans avoir à ouvrir la GUI (Graphical User Interface). La dernière version peut être téléchargée ici : https://github.com/jdemetra/jwsacruncher/releases. Pour plus d’information, vous pouvez visiter la page wiki.
Avec {JDCruncheR}, vous pouvez aussi générer des bilans qualité utilisant l’output du cruncher. Ce bilan est un résumé des diagnostiques de la désaisonnalisation. Il peut être utilisé pour repérer les séries les plus problématiques qui nécessitent une analyse plus fine. Cela est très utile lorsqu’on a beaucoup de séries à désaisonnaliser.
Installation
🎉 {JDCruncheR} est maintenant disponible sur le CRAN ! 🎉
Pour installer, il suffit de lancer la ligne de code suivante :
install.packages("JDCruncheR")
Pour obtenir la version en cours de développement depuis GitHub :
# Si le package remotes n'est pas installé
# install.packages("remotes")
# Installer la version en cours de développement depuis GitHub
remotes::install_github("InseeFr/JDCruncheR")
Usage
Changer les seuils des tests statistiques
Les seuils des tests du bilan qualité sont personnalisables. Pour cela, il faut modifier l’option "jdc_thresholds"
.
Pour récupérer les valeurs des tests par défault, il faut appeler la fonction get_thresholds()
:
get_thresholds("m7", default = TRUE)
#> Good Bad Severe
#> 1 2 Inf
get_thresholds(default = TRUE)
#> $qs_residual_sa_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $qs_residual_sa_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_sa_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_sa_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_td_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_td_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $residuals_independency
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_homoskedasticity
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_skewness
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_kurtosis
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_normality
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $oos_mean
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $oos_mse
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $m7
#> Good Bad Severe
#> 1 2 Inf
#>
#> $q
#> Good Bad
#> 1 Inf
#>
#> $q_m2
#> Good Bad
#> 1 Inf
#>
#> $pct_outliers
#> Good Uncertain Bad
#> 3 5 Inf
#>
#> $grade
#> Good Uncertain Bad Severe
#> 0 1 3 5
Pour changer la valeur de l’option, on peut utiliser la fonction set_thresholds()
:
# Fixer les seuils à une certaine valeur
set_thresholds(test_name = "m7", thresholds = c(Good = 0.8, Bad = 1.4, Severe = Inf))
get_thresholds(test_name = "m7", default = FALSE)
#> Good Bad Severe
#> 0.8 1.4 Inf
# Remettre tous les seuils à leur valeur par défaut
set_thresholds()
get_thresholds(test_name = "m7", default = FALSE)
#> Good Bad Severe
#> 1 2 Inf
Changer les notes des modalités Good
, Uncertain
, Bad
et Severe
Le mécanisme est le même que pour les seuils des tests statistiques avec la valeur "grade"
:
Pour récupérer la valeur par défault des notes, il faut appeler la fonction get_thresholds()
:
get_thresholds("grade", default = TRUE)
#> Good Uncertain Bad Severe
#> 0 1 3 5
Pour changer la valeur de la note, on peut utiliser la fonction set_thresholds()
:
# Fixer les notes à une certaine valeur
set_thresholds(test_name = "grade", thresholds = c(Good = 0, Uncertain = 0.1, Bad = 1, Severe = 10))
get_thresholds(test_name = "grade", default = FALSE)
#> Good Uncertain Bad Severe
#> 0.0 0.1 1.0 10.0
Calculer un bilan qualité
Par exemple, en partant d’une matrice demetra_m.csv
:
n | start | end | mean | skewness | kurtosis | lb2 | p | d | q | bp | bd | bq | m7 | q | q.m2 | ||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
France | 88 | 2012-10-01 | 2020-01-01 | 0.6 | 0.0 | 0.9 | 2.9 | 0.8 | 36.1 | 0.0 | 0 | 1 | 1 | 0 | 1 | 1 | 0.2 | 0.5 | 2.0 |
Spain | 78 | 2015-10-01 | 2022-03-01 | 0.4 | -0.4 | 0.0 | 4.6 | 0.0 | 17.3 | 0.7 | 0 | 0 | 1 | 0 | 1 | 1 | 0.8 | 1.5 | 1.3 |
Greece | 112 | 2010-10-01 | 2020-01-01 | 0.5 | -0.3 | 0.0 | 3.7 | 0.0 | 46.9 | 0.0 | 3 | 1 | 1 | 0 | 1 | 1 | 0.3 | 0.4 | 0.8 |
On peut générer un bilan qualité :
BQ <- extract_QR(x = demetra_m)
print(BQ$modalities)
#> series residuals_homoskedasticity residuals_skewness residuals_kurtosis
#> 1 France Good Good Good
#> 2 Spain Bad Bad Bad
#> 3 Greece Bad Bad Bad
#> oos_mean oos_mse m7 q q_m2 pct_outliers
#> 1 Good <NA> Good Good Bad <NA>
#> 2 Good <NA> Good Bad Bad <NA>
#> 3 Good <NA> Good Bad Good <NA>
Calculer un score
Il est possible maintenant de calculer un score à partir du bilan qualité
BQ_score <- compute_score(
x = BQ,
score_pond = c(
oos_mean = 15L,
residuals_kurtosis = 15L,
residuals_homoskedasticity = 5L,
residuals_skewness = 5L,
m7 = 5L,
q_m2 = 5L
)
)
extract_score(x = BQ_score)
#> series score
#> 1 France 60
#> 2 Spain 110
#> 3 Greece 100
Autres informations
Pour plus d’informations sur l’installation et la configuration du package {JDCruncheR}, vous pouvez visiter la page wiki
Pour une description plus complète des packages R pour JDemetra+ voir le document de travail Insee Les packages R pour JDemetra+ : une aide à la désaisonnalisation
Overview
The primary objective of the {JDCruncheR} package is to provide a quick and easy access to the JDemetra+ cruncher (JWSACruncher
) from R. The cruncher is a tool for updating JDemetra+ workspaces, without having to open the graphical user interface. The latest version can be downloaded here: https://github.com/jdemetra/jwsacruncher/releases. For more information, please refer to the wiki page.
With {JDCruncheR}, you can also generate a quality report based on the cruncher’s output. This report is a formatted summary of the seasonal adjustment process master diagnostics and parameters. It can be used to spot the most problematic series which will require a finer analysis. This is most useful when dealing with a large number of series.
Installation
🎉 {JDCruncheR} is now available on CRAN! 🎉
To install it, you have to launch the following command line:
install.packages("JDCruncheR")
To get the current development version from GitHub:
# If remotes packages is not installed
# install.packages("remotes")
# Install development version from GitHub
remotes::install_github("InseeFr/JDCruncheR")
Usage
Changing statistical test thresholds
The thresholds of the QR tests can be customised You have to modify the option "jdc_thresholds"
.
To get the (default or not) values of the thresholds of the tests, you can call the fonction get_thresholds()
:
get_thresholds("m7")
#> Good Bad Severe
#> 1 2 Inf
get_thresholds(default = TRUE)
#> $qs_residual_sa_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $qs_residual_sa_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_sa_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_sa_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_td_on_sa
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $f_residual_td_on_i
#> Severe Bad Uncertain Good
#> 0.001 0.010 0.050 Inf
#>
#> $residuals_independency
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_homoskedasticity
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_skewness
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_kurtosis
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $residuals_normality
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $oos_mean
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $oos_mse
#> Bad Uncertain Good
#> 0.01 0.10 Inf
#>
#> $m7
#> Good Bad Severe
#> 1 2 Inf
#>
#> $q
#> Good Bad
#> 1 Inf
#>
#> $q_m2
#> Good Bad
#> 1 Inf
#>
#> $pct_outliers
#> Good Uncertain Bad
#> 3 5 Inf
#>
#> $grade
#> Good Uncertain Bad Severe
#> 0 1 3 5
To change the value of the option, you can use the fonction set_thresholds()
:
# Set threshold to imposed value
set_thresholds(test_name = "m7", thresholds = c(Good = 0.8, Bad = 1.4, Severe = Inf))
get_thresholds(test_name = "m7", default = FALSE)
#> Good Bad Severe
#> 0.8 1.4 Inf
# Reset all thresholds to default
set_thresholds()
get_thresholds(test_name = "m7", default = FALSE)
#> Good Bad Severe
#> 1 2 Inf
Changing the scores for the Good
, Uncertain
, Bad
and Severe
modalities
The mechanism is the same as for the statistical test thresholds with the "grade"
value:
To retrieve the default grade value, call the get_thresholds()
function:
get_thresholds("grade", default = TRUE)
#> Good Uncertain Bad Severe
#> 0 1 3 5
To change the value of the grade, you can use the set_thresholds()
function:
# Set grades to a certain value
set_thresholds(test_name = "grade", thresholds = c(Good = 0, Uncertain = 0.1, Bad = 1, Severe = 10))
get_thresholds(test_name = "grade", default = FALSE)
#> Good Uncertain Bad Severe
#> 0.0 0.1 1.0 10.0
Calculate a quality report
For example, starting from a matrix demetra_m.csv
:
n | start | end | mean | skewness | kurtosis | lb2 | p | d | q | bp | bd | bq | m7 | q | q.m2 | ||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
France | 88 | 2012-10-01 | 2020-01-01 | 0.6 | 0.0 | 0.9 | 2.9 | 0.8 | 36.1 | 0.0 | 0 | 1 | 1 | 0 | 1 | 1 | 0.2 | 0.5 | 2.0 |
Spain | 78 | 2015-10-01 | 2022-03-01 | 0.4 | -0.4 | 0.0 | 4.6 | 0.0 | 17.3 | 0.7 | 0 | 0 | 1 | 0 | 1 | 1 | 0.8 | 1.5 | 1.3 |
Greece | 112 | 2010-10-01 | 2020-01-01 | 0.5 | -0.3 | 0.0 | 3.7 | 0.0 | 46.9 | 0.0 | 3 | 1 | 1 | 0 | 1 | 1 | 0.3 | 0.4 | 0.8 |
A quality report can be generated:
BQ <- extract_QR(x = demetra_m)
print(BQ$modalities)
#> series residuals_homoskedasticity residuals_skewness residuals_kurtosis
#> 1 France Good Good Good
#> 2 Spain Bad Bad Bad
#> 3 Greece Bad Bad Bad
#> oos_mean oos_mse m7 q q_m2 pct_outliers
#> 1 Good <NA> Good Good Bad <NA>
#> 2 Good <NA> Good Bad Bad <NA>
#> 3 Good <NA> Good Bad Good <NA>
Calculate a score
It is now possible to calculate a score from the quality report:
BQ_score <- compute_score(
x = BQ,
score_pond = c(
oos_mean = 15L,
residuals_kurtosis = 15L,
residuals_homoskedasticity = 5L,
residuals_skewness = 5L,
m7 = 5L,
q_m2 = 5L
)
)
extract_score(x = BQ_score)
#> series score
#> 1 France 60
#> 2 Spain 110
#> 3 Greece 100
Other informations
For more informations on installing and configuring the {JDCruncheR} package, you can visit the wiki page.
For a more comprehensive description of the R packages for JDemetra+ check the Insee working paper R Tools for JDemetra+: Seasonal adjustment made easier