Package etable.

Introduction

The main purpose of this package is to create descriptive tables for various subgroups in a quick and easy way. Most of the statistics can also be calculated using weights. Below are some examples of packet functionality using artificial data.

Simple tables

With the predifined cell-function iqr_cell one can generate a simple table with interquartile range of a variable. It calculates the median, Q1 and Q3 for bmi variable in data.frame d. The variable to be analysed is selected by setting x_vars='bmi'. With rows='sex' the factor to separate the table by rows is selected. Parameter rnames='Sex' set the label for the row groups.

tab <- tabular.ade(x_vars='bmi', rows='sex', rnames='Sex', data=d, FUN=iqr_cell)
knitr::kable(tab, caption='Median (Q1/Q3) of BMI')

Median (Q1/Q3) of BMI
1	2
Sex
Men	31.2 (27.6/35.2)
Women	31.4 (27.8/35.2)

A simple 2 x 2 table

For a simple 2 x 2 table, the second separation factor for columns needs to be specified, what is done with cols='ethnic' and cnames='Ethnicity'.

tab<-tabular.ade(x_vars='bmi', rows='sex', rnames='Sex', cols='ethnic', cnames='Ethnicity', data=d, FUN=iqr_cell)
knitr::kable(tab, caption='Median (Q1/Q3) of BMI')

Median (Q1/Q3) of BMI
1	2	3	4
	Sex
Ethnicity		Other	Caucasian
	Men	31.2 (27.6/35.2)	31.3 (27.6/35.2)
	Women	31.5 (27.8/35.2)	31.3 (27.9/35.2)

A n(nested) x 2 table

More than one factor at once can be used for rows or columns to create nested tables using rows=c('sex', 'dec'), rnames=c('Sex', 'Decades').

tab<-tabular.ade(x_vars='bmi', rows=c('sex', 'dec'), rnames=c('Sex', 'Decades'),cols='ethnic', cnames='Ethnicity', data=d, FUN=iqr_cell)
knitr::kable(tab, caption='Median (Q1/Q3) of BMI')

Median (Q1/Q3) of BMI
1	2	3	4	5
	Sex	Decades
Ethnicity			Other	Caucasian
	Men	(20,30]	31.5 (28.1/35.5)	30.9 (27.5/35.1)
		(30,40]	31.4 (27.4/35.3)	31.3 (27.5/34.9)
		(40,50]	31.1 (27.7/35.4)	31.1 (27.9/35.1)
		(50,60]	31.9 (27.7/35.2)	31.3 (27.7/34.9)
		(60,70]	31.2 (27.6/35.1)	31.4 (27.5/35.7)
		(70,80]	30.1 (26.7/34.4)	31.4 (27.7/35.2)
	Women	(20,30]	30.9 (26.8/35.8)	31.1 (27.6/35.8)
		(30,40]	31.6 (27.5/34.9)	31.1 (27.7/34.8)
		(40,50]	31.9 (27.8/36.0)	31.6 (28.0/35.5)
		(50,60]	31.0 (27.9/34.7)	31.3 (28.2/35.5)
		(60,70]	31.8 (28.2/35.0)	31.4 (27.8/34.9)
		(70,80]	31.7 (28.8/35.6)	31.2 (27.9/34.6)

A n x n nested table

The cell function n_cell returns the number of non-missing observations in each cell. Missing values of x_vars variable will be excluded.

tab<-tabular.ade(x_vars='sex', rows=c('dec','bmi_q'),  rnames=c('Decades','BMI Quantiles'), cols=c('sex', 'ethnic'), cnames=c('Sex', 'Ethnicity'), data=d, FUN=n_cell)
knitr::kable(tab, caption='N of Obs.')

N of Obs.
1	2	3	4	5	6	7
	Decades	BMI Quantiles
Sex			Men		Women
Ethnicity			Other	Caucasian	Other	Caucasian
	(20,30]	(14.8,27.7]	47	173	64	155
		(27.7,31.3]	64	167	45	150
		(31.3,35.2]	57	150	42	125
		(35.2,65.2]	58	162	53	164
	(30,40]	(14.8,27.7]	50	172	57	157
		(27.7,31.3]	37	144	48	169
		(31.3,35.2]	48	164	67	161
		(35.2,65.2]	46	149	50	146
	(40,50]	(14.8,27.7]	63	156	53	145
		(27.7,31.3]	65	180	42	158
		(31.3,35.2]	55	153	56	157
		(35.2,65.2]	64	161	63	173
	(50,60]	(14.8,27.7]	55	137	47	144
		(27.7,31.3]	48	139	55	171
		(31.3,35.2]	63	142	57	152
		(35.2,65.2]	56	131	44	162
	(60,70]	(14.8,27.7]	49	167	47	151
		(27.7,31.3]	52	135	55	160
		(31.3,35.2]	46	145	58	177
		(35.2,65.2]	48	168	50	148
	(70,80]	(14.8,27.7]	60	150	39	141
		(27.7,31.3]	50	140	54	160
		(31.3,35.2]	33	154	57	163
		(35.2,65.2]	44	154	56	132

A n x 1 table

With the cell function quantile_cell, quantiles can be calculated. The parameter probs defines which quantile should be calculated.

tab<-tabular.ade(x_vars='bmi', xname='BMI', rows=c('sex','ethnic','disease','treat'), rnames=c('Sex', 'Ethnicity', 'Disease', 'Treatment'), data=d, FUN=quantile_cell, probs=0.95)
knitr::kable(tab, caption='95th quantile of BMI')

95th quantile of BMI
1	2	3	4	5
Sex	Ethnicity	Disease	Treatment
Men	Other	no	no	42.0
			yes	41.9
		yes	no	43.0
			yes	44.8
	Caucasian	no	no	41.4
			yes	42.8
		yes	no	43.2
			yes	42.0
Women	Other	no	no	41.3
			yes	41.7
		yes	no	42.4
			yes	39.0
	Caucasian	no	no	41.8
			yes	41.2
		yes	no	40.4
			yes	43.7

Predefined cell functions

There are several predefined cell functions in this package. See the help pages for more information. The stat_cell function includes a wide range of statistics and is the most usefull cell function of all.

stat_cell basic parameters, digits=3, digits2=1
combi_cell basic parameters, digits=3, style=1
mean_sd_cell basic parameters, digits=3, style=1, nsd=1
iqr_cell basic parameters, digits=3, add_n=FALSE
quantile_cell basic parameters, digits=3, probs=0.5, plabels=FALSE
eventpct_cell basic parameters, digits=1, digits2=0, event=2, type=1
miss_cell basic parameters, digits=0, pct=FALSE, prefix=““, suffix=””
corr_p_cell basic parameters, digits=3
mode_cell basic parameters, digits=3

Basic parameters are x, y, z, w, cell_ids, row_ids, col_ids, vnames, vars, n_min. Each cell function must take these parameters. They will be automatically passed from tabular.ade function. Most of the functions use only the x variable for calculations and w for weighted calculations. Only corr_p_cell uses y variable. Additional parameters like digits = 3 can be used in tabular.ade( , ...) instead of the points.

Writing custom cell function

There is a possibility to write custom cell function. It allows all possible designs of the cell and much more.

An example of a custom cell function

my_cell<- function(x, y, z, w, cell_ids, row_ids, col_ids, vnames, vars, n_min)
{
out<- format(mean(x[cell_ids], na.rm=TRUE), digits = 3)
return(out)
}

tab<-tabular.ade(x_vars='age', rows='sex', rnames='Sex', cols='dec', cnames='Decades', data=d, FUN=my_cell)
knitr::kable(tab, caption='Mean Age')

Mean Age
1	2	3	4	5	6	7	8
	Sex
Decades		(20,30]	(30,40]	(40,50]	(50,60]	(60,70]	(70,80]
	Men	25.4	35.6	45.5	55.6	65.7	75.5
	Women	25.4	35.4	45.4	55.2	65.5	75.2

Another simple example of custom cell function

my_cell<- function(x, y, z, w, cell_ids, row_ids, col_ids, vnames, vars, n_min)
{
out<- NULL
tab<-table(x[cell_ids])
for(i in 1:length(tab)){
out<- paste(out, levels(x)[i],': ' ,tab[i], sep='')
if(i<length(tab)) out<- paste(out, ', ', sep='')
}
return(out)
}

tab<-tabular.ade(x_vars='sex', rows='dec', rnames='Decades', cols='stage', cnames='Stage', data=d, FUN=my_cell)
knitr::kable(tab, caption='Frequencies')

Frequencies
1	2	3	4	5
	Decades
Stage		1	2	3
	(20,30]	Men: 444, Women: 408	Men: 341, Women: 307	Men: 93, Women: 83
	(30,40]	Men: 404, Women: 418	Men: 341, Women: 357	Men: 65, Women: 81
	(40,50]	Men: 441, Women: 432	Men: 373, Women: 331	Men: 83, Women: 84
	(50,60]	Men: 369, Women: 427	Men: 330, Women: 330	Men: 72, Women: 75
	(60,70]	Men: 400, Women: 426	Men: 323, Women: 339	Men: 87, Women: 81
	(70,80]	Men: 391, Women: 365	Men: 321, Women: 347	Men: 73, Women: 90

More complicated cell function example

b_cell<- function(x, y, z, w, cell_ids, row_ids, col_ids, vnames, vars, n_min)
{
out<- NULL
if(length(unique(x))==2){
lv<-levels(x)
n <-sum(x[cell_ids]==lv[2])
N <-sum(table(x[cell_ids]))
out<-paste(levels(x)[2], ': ',format((n/N)*100, digits=3),'% (N:',n , ')',sep='')
}
if(!is.factor(x) & length(unique(x))> 2){
quant <- format(quantile(x[cell_ids], c(0.25, 0.5, 0.75), na.rm=TRUE), digits=3)
out<- paste(quant[1], ' (',quant[2],'/',quant[3],')', sep='')
}
if(is.factor(x) & length(unique(x))> 2){
lv<-levels(x)
n <-table(x[cell_ids])
N <-sum(table(x[cell_ids]))
out<- paste(lv, ': ', format((n/N)*100,  digits=3), '%', collapse=' | ', sep='')
}
return(out)
}

tab<-tabular.ade(x_vars=c('bmi','ethnic','stage'),xname=c('BMI','Ethnicity','Stages'), cols='sex', cnames='Sex', data=d, FUN=b_cell)
knitr::kable(tab, caption='Diverse variables')

Diverse variables
1	2	3	4
Sex		Men	Women
	BMI	27.6 (31.2/35.2)	27.8 (31.4/35.2)
	Ethnicity	Caucasian: 74.5% (N:3713)	Caucasian: 74.8% (N:3754)
	Stages	1: 49.41% \| 2: 41.04% \| 3: 9.56%	1: 49.67% \| 2: 40.45% \| 3: 9.88%

A T-test function, usage of x and y variables

t_test_cell<- function(x, y, z, w, cell_ids, row_ids, col_ids, vnames, vars, n_min)
{
v <- x[cell_ids]
group <- y[cell_ids]
test<-t.test(v[which(group==levels(group)[1])], v[which(group==levels(group)[2])])
mdiff<- format(diff(test$estimate), digits=3)
p<- base:::format.pval(test$p.value, digits=2, eps=0.0001)
out<- paste('Diff: ', mdiff, ', p-value: ', p, sep='')
return(out)
}

tab<-tabular.ade(x_vars='bmi', xname='BMI', y_vars='ethnic', yname='Ethnicity', rows='dec', rnames='Decades', cols='sex', cnames='Sex', data=d, FUN=t_test_cell)
knitr::kable(tab, caption='T-test for BMI between Ethnicity groups')

T-test for BMI between Ethnicity groups
1	2	3	4
	Decades
Sex		Men	Women
	(20,30]	Diff: -0.483, p-value: 0.24	Diff: 0.302, p-value: 0.53
	(30,40]	Diff: -0.194, p-value: 0.71	Diff: -0.305, p-value: 0.47
	(40,50]	Diff: 0.171, p-value: 0.7	Diff: -0.0821, p-value: 0.86
	(50,60]	Diff: -0.442, p-value: 0.33	Diff: 0.185, p-value: 0.69
	(60,70]	Diff: -0.15, p-value: 0.76	Diff: -0.0931, p-value: 0.83
	(70,80]	Diff: 0.769, p-value: 0.13	Diff: -0.594, p-value: 0.19

Multiple x or y variables

There is a possibility to pass more than one variable to x_vars or x_vars parameters. In this way a correlation matrix can be created.

vars    <-c('age', 'weight', 'height', 'bmi')
vlabels <-c('Age', 'Weight', 'Height', 'BMI')

tab<-tabular.ade(x_vars=vars, xname=vlabels, y_vars=vars, yname=vlabels,data=d, FUN=corr_p_cell, digits=2)
knitr::kable(tab, caption='Pearson correlation')

Pearson correlation
1	2	3	4	5
	Age	Weight	Height	BMI
Age	1.00	0.01	0.00	0.01
Weight	0.01	1.00	0.01	0.69
Height	0.00	0.01	1.00	-0.71
BMI	0.01	0.69	-0.71	1.00

Multiple x with nested columns

If there are multiple x variables, then they are listed line by line.

vars    <-c('age', 'weight', 'height', 'bmi')
vlabels <-c('Age', 'Weight', 'Height', 'BMI')

tab<-tabular.ade(x_vars=vars, xname=vlabels, cols=c('sex','stage'), cnames=c('Sex','Stage'), data=d, FUN=quantile_cell)
knitr::kable(tab, caption='Medians')

Medians
1	2	3	4	5	6	7	8
Sex		Men			Women
Stage		1	2	3	1	2	3
	Age	49.0	49.0	50.0	50.0	51.0	50.0
	Weight	80.1	79.7	80.4	80.2	80.5	80.5
	Height	1.60	1.60	1.59	1.60	1.60	1.60
	BMI	31.2	31.3	31.4	31.3	31.4	31.5

Complex tables

The `ALL` keyword

The keyword ALL, after a factor in rows or cols statement, adds a row for overall sample.

tab<-tabular.ade(x_vars='sex', rows=c('treat', 'ALL'), rnames=c('Treatment'), cols=c('disease', 'ALL'), cnames=c('Disease'), data=d, FUN=n_cell, alllabel='both')
knitr::kable(tab, caption='Contingency table')

Contingency table
1	2	3	4	5
	Treatment
Disease		no	yes	both
	no	7159	833	7992
	yes	1798	210	2008
	both	8957	1043	10000

Weighted tables

Most of the predefined cell functions support weighting with w=weights parameter. This way weighted statistics can be calculated.

tab<-tabular.ade(x_vars='sex', rows=c('sex', 'ALL', 'ethnic', 'stage'), rnames=c('Sex','Ethnicity', 'Stage'), w='ws', data=d, FUN=n_cell, digits=1)
knitr::kable(tab, caption='weighted N')

weighted N
1	2	3	4
Sex	Ethnicity	Stage
Men	Other	1	486.1
		2	408.0
		3	95.8
	Caucasian	1	1459.6
		2	1224.7
		3	283.6
Women	Other	1	504.5
		2	413.3
		3	107.0
	Caucasian	1	1504.3
		2	1212.1
		3	295.2
Total	Other	1	990.5
		2	821.2
		3	202.7
	Caucasian	1	2963.9
		2	2436.7
		3	578.8

Various statistics in a table

The predefined cell functions stat_cell can calculate several statistics at once. The statistics are set using keywords in x_vars or y_vars parameters.

vars    <-c('age', 'weight', 'height', 'bmi')
vlabels <-c('Age', 'Weight', 'Height', 'BMI')

keywords  <-c('MIN', 'MAX', 'MEAN', 'SD', 'CV', 'SKEW',     'KURT')
keylabels <-c('Min', 'Max', 'Mean', 'SD', 'CV', 'Skewness', 'Kurtosis')

tab<-tabular.ade(x_vars=vars, xname=vlabels, y_vars=keywords, yname=keylabels, data=d, FUN=stat_cell)
knitr::kable(tab, caption='Various statistics')

Various statistics
1	2	3	4	5	6	7	8
	Min	Max	Mean	SD	CV	Skewness	Kurtosis
Age	20.0	80.0	49.9	17.4	0.348	0.0150	-1.20
Weight	35.2	115	80.0	10.0	0.125	-0.0475	0.0314
Height	1.19	2.02	1.60	0.101	0.0630	0.0107	0.0500
BMI	14.8	65.2	31.7	5.69	0.180	0.484	0.671

Various statistics combined with `rows` parameter

keywords  <-c('N', 'MIN', 'MAX', 'MEAN', 'SD')
keylabels <-c('N', 'Min', 'Max', 'Mean', 'SD')

tab<-tabular.ade(x_vars=vars, xname=vlabels, y_vars=keywords, yname=keylabels, rows=c('sex','ALL','ethnic'), rnames=c('Sex','Ethnicity'), data=d, FUN=stat_cell)
knitr::kable(tab, caption='Various statistics')

Various statistics
1	2	3	4	5	6	7	8
	Sex	Ethnicity
			N	Min	Max	Mean	SD
Age	Men	Other	1268	20.0	80.0	49.7	17.2
		Caucasian	3713	20.0	80.0	49.8	17.6
	Women	Other	1265	20.0	80.0	50.1	17.3
		Caucasian	3754	20.0	80.0	50.1	17.2
	Total	Other	2533	20.0	80.0	49.9	17.3
		Caucasian	7467	20.0	80.0	50.0	17.4
Weight	Men	Other	1268	44.2	110	80.0	9.98
		Caucasian	3713	41.1	114	79.8	10.1
	Women	Other	1265	43.1	111	80.5	10.0
		Caucasian	3754	35.2	115	80.0	9.97
	Total	Other	2533	43.1	111	80.2	10.0
		Caucasian	7467	35.2	115	79.9	10.0
Height	Men	Other	1268	1.19	1.98	1.60	0.103
		Caucasian	3713	1.23	2.02	1.60	0.101
	Women	Other	1265	1.24	1.98	1.60	0.0989
		Caucasian	3754	1.21	1.95	1.60	0.101
	Total	Other	2533	1.19	1.98	1.60	0.101
		Caucasian	7467	1.21	2.02	1.60	0.101
BMI	Men	Other	1268	15.0	58.7	31.7	5.88
		Caucasian	3713	15.8	65.2	31.6	5.68
	Women	Other	1265	15.5	64.9	31.8	5.64
		Caucasian	3754	14.8	58.6	31.7	5.66
	Total	Other	2533	15.0	64.9	31.7	5.76
		Caucasian	7467	14.8	65.2	31.7	5.67

An example using the statistic keywords in `x_vars` parameter

keywords  <-c('N', 'MIN', 'MAX', 'MEAN', 'SD')
keylabels <-c('N', 'Min', 'Max', 'Mean', 'SD')


tab<-tabular.ade(x_vars=keywords, xname=keylabels, y_vars=vars, yname=vlabels, rows=c('sex', 'ALL'), rnames=c('Sex'),data=d, FUN=stat_cell)
knitr::kable(tab, caption='Various statistics')

Various statistics
1	2	3	4	5	6
	Sex
		Age	Weight	Height	BMI
N	Men	4981	4981	4981	4981
	Women	5019	5019	5019	5019
	Total	10000	10000	10000	10000
Min	Men	20.0	41.1	1.19	15.0
	Women	20.0	35.2	1.21	14.8
	Total	20.0	35.2	1.19	14.8
Max	Men	80.0	114	2.02	65.2
	Women	80.0	115	1.98	64.9
	Total	80.0	115	2.02	65.2
Mean	Men	49.8	79.9	1.60	31.6
	Women	50.1	80.1	1.60	31.7
	Total	49.9	80.0	1.60	31.7
SD	Men	17.5	10.0	0.101	5.73
	Women	17.3	9.98	0.100	5.66
	Total	17.4	10.0	0.101	5.69

And finally, an example of a weighted, multivariable, nested table with several statistics

vars    <-c('age', 'weight', 'height', 'bmi')
vlabels <-c('Age', 'Weight', 'Height', 'BMI')

keywords  <-c('N', 'MEDIAN', 'IQR')
keylabels <-c('N', 'Median', 'IQR')

tab<-tabular.ade(x_vars=vars, xname=vlabels, y_vars=keywords, yname=keylabels, rows=c('sex', 'ALL'), rnames=c('Sex'),cols=c('ethnic'),cnames=c('Ethnicity'),w='ws',data=d,FUN=stat_cell)
knitr::kable(tab, caption='Various statistics')

Various statistics
1	2	3	4	5	6	7	8	9
		Sex
			N		Median		IQR
Ethnicity			Other	Caucasian	Other	Caucasian	Other	Caucasian
	Age	Men	990	2968	49.0	49.0	29.0	31.0
		Women	1025	3012	50.0	50.0	30.0	29.0
		Total	2015	5979	50.0	49.0	30.0	30.0
	Weight	Men	990	2968	80.3	80.1	13.2	13.3
		Women	1025	3012	80.6	80.2	12.9	13.3
		Total	2015	5979	80.5	80.2	13.2	13.4
	Height	Men	990	2968	1.60	1.60	0.140	0.137
		Women	1025	3012	1.60	1.60	0.122	0.139
		Total	2015	5979	1.60	1.60	0.132	0.138
	BMI	Men	990	2968	31.2	31.3	7.99	7.50
		Women	1025	3012	31.4	31.3	7.53	7.18
		Total	2015	5979	31.3	31.3	7.77	7.35