form.io - Nicely formatted objects output in R Markdown

Philippe Grosjean (phgrosjean@sciviews.org)

2020-04-27

Among the ‘SciViews’ R packages, ‘form.io’ would be devoted to the input and especially output of textual data (input of tabular data being managed by ‘data.io’). Here, we reuse existing functions like knitr::kable() or pander::pander(), but we make them simpler to use and working in all cases being:

• At the R console,
• Inside an R Markdown/Notebook inline area just beneath a chunk,
• In a knitted R markdown file, HTML format,
• Idem, but with a LaTeX or Word format.

Both ‘knitr’ with kable(), and the ‘kableExtra’ functions, and ‘pander’ with the pander() functions do that. However, the syntax is not always easy, and code eventually differ depending on the situation among the four possible ones here above.

Moreover, none of kable() or pander() functions allow for the translation of the result in a different language, and they also do not take into account attributes like label or units (see the ‘data.io’ package).

So, the new generic function form() creates formatted tables or other objects (lists, …) to be directly integrated in an R Markdown document, or to be used at the R Console. Just load the package (and perhaps also ‘data.io’, or even the whole SciViews::R suite):

# Better to use SciViews::R() here
library(data.io)
#> 
#> Attaching package: 'data.io'
#> The following object is masked from 'package:base':
#> 
#>     write
library(form.io)
#> Registered S3 method overwritten by 'svMisc':
#>   method             from   
#>   $.subsettable_type data.io
# To change language to, e.g., French
options(data.io_lang = "FR") # Same language option is used by form.io::form()

# Test
trees <- read("trees", package = "datasets")
lm_1 <- lm(data = trees, volume ~ .)
form(trees)

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56
0,28	20	0,44
0,28	23	0,52
0,28	24	0,64
0,28	23	0,56
0,29	24	0,69
0,29	23	0,59
0,29	23	0,61
0,3	21	0,6
0,3	23	0,54
0,33	23	0,63
0,33	26	0,96
0,34	26	0,78
0,35	22	0,73
0,35	20	0,7
0,36	24	0,98
0,36	24	0,9
0,37	23	1
0,41	22	1,1
0,41	24	1,2
0,44	25	1,6
0,44	25	1,6
0,46	24	1,7
0,46	24	1,5
0,46	24	1,4
0,52	26	2,2

form(head(trees), data = trees) # Try without data =

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56

form(anova(lm_1), data = trees)

Analyse de la variance - Réponse : Volume de bois

	DDL	Somme carrés	Carrés moyens	Stat. F	Valeur p
Diamètre à 1,4m	1	6,1	6,1	499	2,2e-19	* * *
Hauteur	1	0,081	0,081	6,6	0,016	*
Résidus	28	0,34	0,012

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ’ ’ 1

Using form()

pander::pander() works well with knitted documents, but it outputs text only in R notebooks. The form() function is solving this problem, while still outputting a textual markdown version at the R console.

trees <- read("trees", package = "datasets")

lm_1 <- lm(data = trees, volume ~ .)
lm_2 <- lm( data = trees, volume ~ I(diameter^2))

• Data frame :

form(trees, caption = "Mesure de cerisiers noirs")

Mesure de cerisiers noirs

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56
0,28	20	0,44
0,28	23	0,52
0,28	24	0,64
0,28	23	0,56
0,29	24	0,69
0,29	23	0,59
0,29	23	0,61
0,3	21	0,6
0,3	23	0,54
0,33	23	0,63
0,33	26	0,96
0,34	26	0,78
0,35	22	0,73
0,35	20	0,7
0,36	24	0,98
0,36	24	0,9
0,37	23	1
0,41	22	1,1
0,41	24	1,2
0,44	25	1,6
0,44	25	1,6
0,46	24	1,7
0,46	24	1,5
0,46	24	1,4
0,52	26	2,2

# Equivalent to:
#form(trees, caption = "Mesure de cerisiers noirs", labels = TRUE, units = TRUE)

Any manipulation of a data frame with labels and units makes it loose its attributes, … but the data = argument is there to look at original data set for these attributes :

form(head(trees))

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56

form(head(trees), data = trees)

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56

• Objects data.table :

library(data.table)
trees_dt <- as.data.table(trees)
class(trees_dt)
#> [1] "data.table" "data.frame"
form(trees_dt)

Diamètre à 1,4m [m]	Hauteur [m]	Volume de bois [m3]
0,21	21	0,29
0,22	20	0,29
0,22	19	0,29
0,27	22	0,46
0,27	25	0,53
0,27	25	0,56
0,28	20	0,44
0,28	23	0,52
0,28	24	0,64
0,28	23	0,56
0,29	24	0,69
0,29	23	0,59
0,29	23	0,61
0,3	21	0,6
0,3	23	0,54
0,33	23	0,63
0,33	26	0,96
0,34	26	0,78
0,35	22	0,73
0,35	20	0,7
0,36	24	0,98
0,36	24	0,9
0,37	23	1
0,41	22	1,1
0,41	24	1,2
0,44	25	1,6
0,44	25	1,6
0,46	24	1,7
0,46	24	1,5
0,46	24	1,4
0,52	26	2,2

• Object lm :

form(lm_1, data = trees, signif_stars = TRUE)

Paramètres de la régression linéaire : volume ~ .

	Estimateur	Erreur type	Stat. t	Valeur p
(Ordonnée à l’origine)	-1,6	0,24	-6,7	2,9e-07	* * *
Diamètre à 1,4m	5,3	0,3	18	8,9e-17	* * *
Hauteur	0,031	0,012	2,6	0,016	*

• Summary of lm :

form(summary(lm_1))

Résumé de la régression linéaire : volume ~ .

Observations	Erreur type des résidus	R2	R2 ajusté
31	0,11	0,948	0,944

• Anova of lm :

form(anova(lm_1), data = trees) #, caption = "ANOVA de ma regression linéaire")

Analyse de la variance - Réponse : Volume de bois

	DDL	Somme carrés	Carrés moyens	Stat. F	Valeur p
Diamètre à 1,4m	1	6,1	6,1	499	2,2e-19	* * *
Hauteur	1	0,081	0,081	6,6	0,016	*
Résidus	28	0,34	0,012

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ’ ’ 1

• Generalized linear models :

# From ?glm
## Dobson (1990) Page 93: Randomized Controlled Trial :
ad <- data.frame(treatment = gl(3, 3), outcome = gl(3, 1, 9),
  counts = c(18, 17, 15, 20, 10, 20, 25, 13, 12))
glm_1 <- glm(data = ad, counts ~ outcome + treatment, family = poisson())
glm_1
#> 
#> Call:  glm(formula = counts ~ outcome + treatment, family = poisson(), 
#>     data = ad)
#> 
#> Coefficients:
#> (Intercept)     outcome2     outcome3   treatment2   treatment3  
#>   3.045e+00   -4.543e-01   -2.930e-01    1.338e-15    1.421e-15  
#> 
#> Degrees of Freedom: 8 Total (i.e. Null);  4 Residual
#> Null Deviance:       10.58 
#> Residual Deviance: 5.129     AIC: 56.76
summary(glm_1)
#> 
#> Call:
#> glm(formula = counts ~ outcome + treatment, family = poisson(), 
#>     data = ad)
#> 
#> Deviance Residuals: 
#>        1         2         3         4         5         6         7         8  
#> -0.67125   0.96272  -0.16965  -0.21999  -0.95552   1.04939   0.84715  -0.09167  
#>        9  
#> -0.96656  
#> 
#> Coefficients:
#>               Estimate Std. Error z value Pr(>|z|)    
#> (Intercept)  3.045e+00  1.709e-01  17.815   <2e-16 ***
#> outcome2    -4.543e-01  2.022e-01  -2.247   0.0246 *  
#> outcome3    -2.930e-01  1.927e-01  -1.520   0.1285    
#> treatment2   1.338e-15  2.000e-01   0.000   1.0000    
#> treatment3   1.421e-15  2.000e-01   0.000   1.0000    
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> 
#> (Dispersion parameter for poisson family taken to be 1)
#> 
#>     Null deviance: 10.5814  on 8  degrees of freedom
#> Residual deviance:  5.1291  on 4  degrees of freedom
#> AIC: 56.761
#> 
#> Number of Fisher Scoring iterations: 4
anova(glm_1)
#> Analysis of Deviance Table
#> 
#> Model: poisson, link: log
#> 
#> Response: counts
#> 
#> Terms added sequentially (first to last)
#> 
#> 
#>           Df Deviance Resid. Df Resid. Dev
#> NULL                          8    10.5814
#> outcome    2   5.4523         6     5.1291
#> treatment  2   0.0000         4     5.1291

anorexia <- read("anorexia", package = "MASS")
form(anorexia)

Treat	Prewt	Postwt
Cont	81	80
Cont	89	80
Cont	92	86
Cont	74	86
Cont	78	76
Cont	88	78
Cont	87	75
Cont	75	87
Cont	81	74
Cont	78	85
Cont	78	77
Cont	89	80
Cont	81	90
Cont	78	81
Cont	70	82
Cont	77	77
Cont	85	84
Cont	86	75
Cont	84	80
Cont	80	73
Cont	86	88
Cont	84	85
Cont	80	81
Cont	78	81
Cont	72	88
Cont	89	79
CBT	80	82
CBT	85	86
CBT	82	81
CBT	83	82
CBT	80	76
CBT	89	104
CBT	95	98
CBT	76	93
CBT	81	73
CBT	80	82
CBT	85	97
CBT	89	95
CBT	81	82
CBT	76	72
CBT	70	91
CBT	80	71
CBT	83	85
CBT	83	82
CBT	88	89
CBT	84	84
CBT	86	83
CBT	76	76
CBT	80	83
CBT	88	100
CBT	83	85
CBT	80	84
CBT	84	85
CBT	81	96
CBT	87	87
FT	84	95
FT	83	94
FT	86	92
FT	82	92
FT	87	100
FT	80	77
FT	77	77
FT	94	102
FT	73	95
FT	80	75
FT	82	78
FT	82	96
FT	78	91
FT	84	92
FT	90	94
FT	86	92
FT	87	98

glm_2 <- glm(data = anorexia, Postwt ~ Prewt + Treat, family = gaussian)
glm_3 <- glm(data = anorexia, Postwt ~ Prewt + Treat + offset(Prewt), family = gaussian)
form(glm_2)
#> 
#> ( Dispersion parameter for   gaussian   family taken to be   48.69504 )

Paramètres de la régression linéaire : Postwt ~ Prewt + Treat

Null deviance:	4584 on 71 degrees of freedom
Residual deviance:	3311 on 68 degrees of freedom

form(summary(glm_2))
#> 
#> ( Dispersion parameter for   gaussian   family taken to be   48.69504 )

Paramètres de la régression linéaire généralisée : Postwt ~ Prewt + Treat

Null deviance:	4584 on 71 degrees of freedom
Residual deviance:	3311 on 68 degrees of freedom

#form(anova(glm_2))
form(aov(glm_2))

Paramètres de la régression linéaire : glm_2

	Df	Sum Sq	Mean Sq	F value	Pr(>F)
Prewt	1	507	507	10	0,0019
Treat	2	766	383	7,9	0,00084
Residuals	68	3311	49

# A Gamma example, from McCullagh & Nelder (1989, pp. 300-2)
clotting <- data.frame(
  u = c(5, 10, 15, 20, 30, 40, 60, 80, 100),
  lot1 = c(118, 58, 42, 35, 27, 25, 21, 19, 18),
  lot2 = c(69, 35, 26, 21, 18, 16, 13, 12, 12))
glm_4 <- glm(data = clotting, lot1 ~ log(u), family = Gamma)
glm_4
#> 
#> Call:  glm(formula = lot1 ~ log(u), family = Gamma, data = clotting)
#> 
#> Coefficients:
#> (Intercept)       log(u)  
#>    -0.01655      0.01534  
#> 
#> Degrees of Freedom: 8 Total (i.e. Null);  7 Residual
#> Null Deviance:       3.513 
#> Residual Deviance: 0.01673   AIC: 37.99
summary(glm_4)
#> 
#> Call:
#> glm(formula = lot1 ~ log(u), family = Gamma, data = clotting)
#> 
#> Deviance Residuals: 
#>      Min        1Q    Median        3Q       Max  
#> -0.04008  -0.03756  -0.02637   0.02905   0.08641  
#> 
#> Coefficients:
#>               Estimate Std. Error t value Pr(>|t|)    
#> (Intercept) -0.0165544  0.0009275  -17.85 4.28e-07 ***
#> log(u)       0.0153431  0.0004150   36.98 2.75e-09 ***
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> 
#> (Dispersion parameter for Gamma family taken to be 0.002446059)
#> 
#>     Null deviance: 3.51283  on 8  degrees of freedom
#> Residual deviance: 0.01673  on 7  degrees of freedom
#> AIC: 37.99
#> 
#> Number of Fisher Scoring iterations: 3
anova(glm_4)
#> Analysis of Deviance Table
#> 
#> Model: Gamma, link: inverse
#> 
#> Response: lot1
#> 
#> Terms added sequentially (first to last)
#> 
#> 
#>        Df Deviance Resid. Df Resid. Dev
#> NULL                       8     3.5128
#> log(u)  1   3.4961         7     0.0167

glm_5 <- glm(data = clotting, lot2 ~ log(u), family = Gamma)
glm_5
#> 
#> Call:  glm(formula = lot2 ~ log(u), family = Gamma, data = clotting)
#> 
#> Coefficients:
#> (Intercept)       log(u)  
#>    -0.02391      0.02360  
#> 
#> Degrees of Freedom: 8 Total (i.e. Null);  7 Residual
#> Null Deviance:       3.119 
#> Residual Deviance: 0.01267   AIC: 27.03
summary(glm_5)
#> 
#> Call:
#> glm(formula = lot2 ~ log(u), family = Gamma, data = clotting)
#> 
#> Deviance Residuals: 
#>      Min        1Q    Median        3Q       Max  
#> -0.05574  -0.02925   0.01030   0.01714   0.06371  
#> 
#> Coefficients:
#>               Estimate Std. Error t value Pr(>|t|)    
#> (Intercept) -0.0239085  0.0013265  -18.02 4.00e-07 ***
#> log(u)       0.0235992  0.0005768   40.91 1.36e-09 ***
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> 
#> (Dispersion parameter for Gamma family taken to be 0.001813354)
#> 
#>     Null deviance: 3.118557  on 8  degrees of freedom
#> Residual deviance: 0.012672  on 7  degrees of freedom
#> AIC: 27.032
#> 
#> Number of Fisher Scoring iterations: 3
anova(glm_5)
#> Analysis of Deviance Table
#> 
#> Model: Gamma, link: inverse
#> 
#> Response: lot2
#> 
#> Terms added sequentially (first to last)
#> 
#> 
#>        Df Deviance Resid. Df Resid. Dev
#> NULL                       8    3.11856
#> log(u)  1   3.1059         7    0.01267

## Aliased ("S"ingular) -> 1 NA coefficient
(fS <- glm(lot2 ~ log(u) + log(u^2), data = clotting, family = Gamma))
#> 
#> Call:  glm(formula = lot2 ~ log(u) + log(u^2), family = Gamma, data = clotting)
#> 
#> Coefficients:
#> (Intercept)       log(u)     log(u^2)  
#>    -0.02391      0.02360           NA  
#> 
#> Degrees of Freedom: 8 Total (i.e. Null);  7 Residual
#> Null Deviance:       3.119 
#> Residual Deviance: 0.01267   AIC: 27.03
tools::assertError(update(fS, singular.ok = FALSE), verbose = interactive())
## -> .. "singular fit encountered"

• Nonlinear regression objects :

nls_1 <- nls(data = trees, volume ~ SSlogis(diameter, Asymptote, Inflexion, Echelle))
form(nls_1)

Paramètres de la régression non linéaire : volume ~ SSlogis(diameter, Asymptote, Inflexion, Echelle)

	Estimateur	Erreur type	Stat. t	Valeur p
Asymptote	5	2,1	2,4	0,026
Inflexion	0,56	0,096	5,8	2,8e-06
Echelle	0,13	0,018	7,5	4e-08

form(nls_1, params_labels = c("Asym", "xmoyen", "scal")) # Rename the parameters

Paramètres de la régression non linéaire : volume ~ SSlogis(diameter, Asymptote, Inflexion, Echelle)

	Estimateur	Erreur type	Stat. t	Valeur p
Asym	5	2,1	2,4	0,026
xmoyen	0,56	0,096	5,8	2,8e-06
scal	0,13	0,018	7,5	4e-08

form(summary(nls_1))

Résumé de la régression non linéaire : volume ~ SSlogis(diameter, Asymptote, Inflexion, Echelle)

Observations	DDL	Erreur type des résidus	Itérations	Tolérance finale
31	28	0,0943	3	5,12e-07

nls_2 <- nls(data = trees, volume ~ SSgompertz(diameter, Asymptote, B2, B3))
form(anova(nls_1, nls_2))

DDL rés. | Somme carrés rés. | DDL | Somme carrés | Stat. F | Valeur p | |

|:——–:|:—————–:|:—:|:————:|:———:|:———-:|::| | 28 | 0,25 | | | | | | | 28 | 0,25 | 0 | 0 | | | |

Table: Analyse de la variance - modèle 1: volume ~ SSlogis(diameter, Asymptote, Inflexion, Echelle), modèle 2: volume ~ SSgompertz(diameter, Asymptote, B2, B3)

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ’ ’ 1

• Lists :

form(list(a = 1:3, b = TRUE, c = LETTERS))

• a: 1, 2 et 3
• b: TRUE
• c: A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y et Z

• Objects ts :

form(nottem, caption = "températures à Notthingam")

températures à Notthingam

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
1920	41	41	44	47	54	58	58	56	54	50	43	40
1921	44	40	45	47	54	59	66	60	57	54	40	43
1922	38	39	40	42	56	58	57	54	54	47	42	42
1923	42	40	43	46	49	53	64	60	54	49	36	38
1924	39	38	38	46	53	58	61	58	56	50	44	44
1925	40	40	41	45	54	59	64	61	53	50	38	36
1926	39	43	43	49	51	57	62	62	58	47	42	40
1927	39	38	45	47	52	55	60	60	55	50	42	35
1928	41	41	43	47	51	56	62	60	55	50	43	37
1929	35	31	41	44	53	57	62	60	60	49	43	42
1930	42	37	41	47	51	60	60	62	57	51	43	39
1931	37	38	38	46	54	58	61	58	54	47	46	41
1932	42	38	40	45	51	57	62	64	56	47	44	42
1933	36	39	44	49	54	61	66	65	60	50	42	36
1934	39	38	40	47	53	60	66	60	59	51	43	46
1935	40	43	44	47	50	60	65	64	57	49	44	36
1936	37	35	44	44	53	59	60	61	58	50	42	41
1937	41	41	38	47	54	59	61	62	56	51	41	37
1938	42	41	47	47	52	59	60	60	57	51	48	39
1939	39	41	42	48	52	58	61	62	58	47	47	38

• Unknown object type (coerced to a list with a warning) :

form(structure(list(a = 1, b = 3:5), class = "new_oject"))
#> Warning in pander.default(x, ...): No pander.method for "new_oject", reverting
#> to default.

• a: 1
• b: 3, 4 et 5

Using kable()

Note: the kable() functions are not integrated yet into form() for the moment! The kable() function is standard in ‘knitr’. Moreover, the ‘kableExtra’ package allows for further configuration of the output.

#options( knitr.table.format = "markdown")
knitr::kable(trees, caption = "The `trees` dataset")

The trees dataset

diameter	height	volume
0.211	21.3	0.292
0.218	19.8	0.292
0.224	19.2	0.289
0.267	21.9	0.464
0.272	24.7	0.532
0.274	25.3	0.558
0.279	20.1	0.442
0.279	22.9	0.515
0.282	24.4	0.640
0.284	22.9	0.563
0.287	24.1	0.685
0.290	23.2	0.595
0.290	23.2	0.606
0.297	21.0	0.603
0.305	22.9	0.541
0.328	22.6	0.629
0.328	25.9	0.957
0.338	26.2	0.776
0.348	21.6	0.728
0.351	19.5	0.705
0.356	23.8	0.977
0.361	24.4	0.898
0.368	22.6	1.028
0.406	21.9	1.085
0.414	23.5	1.206
0.439	24.7	1.569
0.444	25.0	1.577
0.455	24.4	1.651
0.457	24.4	1.458
0.457	24.4	1.444
0.523	26.5	2.180

#kableExtra::xtable2kable(xtable::xtable(linear.1))

knitr::kable(broom::glance(lm_1)[, c(1, 3, 5, 8, 11)])

r.squared	sigma	p.value	AIC	df.residual
0.947516	0.1103626	0	-43.82811	28

knitr::kable(broom::tidy(lm_1), caption = "Tableau des coefficients")

Tableau des coefficients

term	estimate	std.error	statistic	p.value
(Intercept)	-1.6356345	0.2446222	-6.686369	0.0000003
diameter	5.2564322	0.2959381	17.761931	0.0000000
height	0.0311215	0.0120890	2.574370	0.0156218

Using stargazer

Not good! But we may considering to fix these…

#stargazer::stargazer(trees, type = "text")
#stargazer::stargazer(trees, type = "latex")
#stargazer::stargazer(lm_1, type = "text")