| Title: | T Loux Doing R: Functions to Simplify Data Analysis and Reporting |
|---|---|
| Description: | Gives a number of functions to aid common data analysis processes and reporting statistical results in an 'RMarkdown' file. Data analysis functions combine multiple base R functions used to describe simple bivariate relationships into a single, easy to use function. Reporting functions will return character strings to report p-values, confidence intervals, and hypothesis test and regression results. Strings will be LaTeX-formatted as necessary and will knit pretty in an 'RMarkdown' document. The package also provides wrappers function in the 'tableone' package to make the results knit-able. |
| Authors: | Travis Loux [aut, cre] |
| Maintainer: | Travis Loux <[email protected]> |
| License: | GPL-3 |
| Version: | 0.4.0 |
| Built: | 2025-02-14 06:53:23 UTC |
| Source: | CRAN |
as_perc formats a proportion as a percentage to print in an RMarkdown
document
as_perc(p, digits = 0)as_perc(p, digits = 0)
p |
A length-1 numeric to be interpreted as a proportion |
digits |
Number of digits to round percentage to (default to 0) |
Simply multiplies p by 100 and affixes a percent sign to the end after
rounding.
Returns a string to report a percentage to the specified number of digits.
as_perc(0.2345) as_perc(0.000234)as_perc(0.2345) as_perc(0.000234)
cat_compare gives details about the association between two categorical variables.
cat_compare(x, y, plot = TRUE)cat_compare(x, y, plot = TRUE)
x |
A categorical variable: the predictor or group variable, if appropriate |
y |
A categorical variable: the outcome, if appropriate |
plot |
Logical. Whether a mosaic plot should be drawn |
Strictly, x and y do not need to be factors but will be coerced into factors.
Returns a list including (1) two-way table of counts, (2) chi-squared test for independence, (3) Cramer's V standardized effect, and (4) ggplot2 column plot of proportions conditional on x, if requested.
The table of counts will include missing values of both variables, but these rows/columns are discarded prior to the chi-squared test and Cramer's V calculations.
v1 = rbinom(n=50, size=1, p=0.5) v2 = rbinom(n=50, size=2, p=0.3 + 0.2*v1) cat_compare(x=v1, y=v2, plot=TRUE)v1 = rbinom(n=50, size=1, p=0.5) v2 = rbinom(n=50, size=2, p=0.3 + 0.2*v1) cat_compare(x=v1, y=v2, plot=TRUE)
Deprecated. Use 'num_compare' instead.
cont_compare(y, grp, plot = c("density", "boxplot", "none"))cont_compare(y, grp, plot = c("density", "boxplot", "none"))
y |
A numerical variable |
grp |
A categorical variable |
plot |
Type of plot to produce |
Returns a list including (1) group-wise summary statistics, (2) ANOVA decomposition, (3) eta-squared effect size, and (4) ggplot2 object, if requested.
cutp is a wrapper for the base 'cut' function. The vector 'x' will be categorized using the percentiles provided in 'p' to create break values.
cutp(x, p, ...)cutp(x, p, ...)
x |
A numeric vector to be discretized |
p |
A numeric vector of probabilities |
... |
Arguments passed to 'cut' |
Within the 'cutp' function, 'p' is passed to 'quantile' as the 'probs' input. The computed quantiles are then used as the 'breaks' in 'cut'.
The values '-Inf' and 'Inf' are added to the beginning and end of the breaks vector, respectively, so quantiles for 0 and 1 do not need to be given explicitly.
Returns the output from 'cut'. This is usually a factor unless otherwise specified.
myvals = rnorm(1000) catx = cutp(x=myvals, p=c(0.25, 0.5, 0.75), labels=c('Q1', 'Q2', 'Q3', 'Q4')) table(catx)myvals = rnorm(1000) catx = cutp(x=myvals, p=c(0.25, 0.5, 0.75), labels=c('Q1', 'Q2', 'Q3', 'Q4')) table(catx)
inline_coef presents the resuts of a coefficient from a lm or glm model in LaTeX format to be reported inline in an RMarkdown document.
inline_coef(model, variable, coef = TRUE, stat = TRUE, pval = TRUE, digits = 2) inline_coef_p(model, variable, digits = 2)inline_coef(model, variable, coef = TRUE, stat = TRUE, pval = TRUE, digits = 2) inline_coef_p(model, variable, digits = 2)
model |
A regression model |
variable |
A character string giving the name of the variable to be reported |
coef |
Logical, whether the coefficient value is to be reported (default TRUE) |
stat |
Logical, whether the test statistic for the coefficient should be reported (default TRUE) |
pval |
Logical, whether the p-value for the coefficient should be reported (default TRUE) |
digits |
Number of digits to round to (default to 2) |
This function currently only supports lm and glm objects. Suggestions and requests are welcomed.
inline_coef_p is a wrapper for inline_coef to report only the p-value (sets all non-p-value logicals to FALSE).
Returns a LaTeX-formatted result for use in RMarkdown document.
x1 = rnorm(20) x2 = rnorm(20) y = x1 + x2 + rnorm(20) model1 = lm(y ~ x1 + x2) inline_coef(model1, 'x1') inline_coef_p(model1, 'x1')x1 = rnorm(20) x2 = rnorm(20) y = x1 + x2 + rnorm(20) model1 = lm(y ~ x1 + x2) inline_coef(model1, 'x1') inline_coef_p(model1, 'x1')
inline_reg presents the fit of a coefficient from a lm or glm model in LaTeX format to be reported inline in an RMarkdown document.
inline_reg(model, fit = TRUE, stat = TRUE, pval = TRUE, digits = 2) inline_reg_p(model, digits = 2) inline_anova(model, stat = TRUE, pval = TRUE, digits = 2)inline_reg(model, fit = TRUE, stat = TRUE, pval = TRUE, digits = 2) inline_reg_p(model, digits = 2) inline_anova(model, stat = TRUE, pval = TRUE, digits = 2)
model |
A regression model |
fit |
Logical, whether the regression fit is to be reported (default TRUE, only applicable to |
stat |
Logical, whether the test statistic for the coefficient should be reported (default TRUE) |
pval |
Logical, whether the p-value for the coefficient should be reported (default TRUE) |
digits |
Number of digits to round to (default to 2) |
For lm objects, results include R-squared, the F statistic, and the p-value. For glm objects, results include the chi-squared statistic and the p-value.
This function currently only supports lm and glm objects. Suggestions and requests are welcomed.
inline_reg_p is a wrapper for inline_reg to report only the p-value (sets all non-p-value logicals to FALSE). inline_anova is a wrapper to report a one-way ANOVA result in which fit is set to FALSE and other logical inputs (stat, pval, and digits) are allowed to be user-defined.
Returns a LaTeX-formatted result for use in RMarkdown document.
x1 = rnorm(20) y1 = x1 + rnorm(20) model1 = lm(y1 ~ x1) inline_reg(model1) x2 = rnorm(20) y2 = rbinom(n=20, size=1, prob=pnorm(x2)) model2 = glm(y2 ~ x2, family=binomial('logit')) inline_reg(model2)x1 = rnorm(20) y1 = x1 + rnorm(20) model1 = lm(y1 ~ x1) inline_reg(model1) x2 = rnorm(20) y2 = rbinom(n=20, size=1, prob=pnorm(x2)) model2 = glm(y2 ~ x2, family=binomial('logit')) inline_reg(model2)
inline_test formats the results of an htest object into LaTeX to be presented inline in an RMarkdown document.
inline_test(test, stat = TRUE, pval = TRUE, digits = 2) inline_test_p(test, digits = 2)inline_test(test, stat = TRUE, pval = TRUE, digits = 2) inline_test_p(test, digits = 2)
test |
An htest object |
stat |
Logical, whether to report test statistic (default TRUE) |
pval |
Logical, whether to report p-value (default TRUE) |
digits |
Number of digits to round to (default to 2) |
This function currently only supports t tests and chi-squared tests. Suggestions and requests are welcomed.
inline_test_p is a wrapper for inline_test to report only the p-value (sets all non-p-value logicals to FALSE).
Returns a LaTeX-formatted hypothesis test result for use in RMarkdown document.
x = rnorm(20) test1 = t.test(x) inline_test(test1) inline_test_p(test1)x = rnorm(20) test1 = t.test(x) inline_test(test1) inline_test_p(test1)
KreateTableOne is a wrapper for tableone::CreateTableOne which
formats the original plain text table as a data.frame of character columns.
KnitableTableOne is a wrapper for tableone::print.TableOne which
allows for more versatility in printing options. The output of both functions
can be printed in an RMarkdown document in a number of ways, e.g., using
knitr::kable. svyKreateTabeOne does the same with
tableone::svyCreateTableOne for complex survey data.
KreateTableOne(...) svyKreateTableOne(...) KnitableTableOne(x, ...)KreateTableOne(...) svyKreateTableOne(...) KnitableTableOne(x, ...)
... |
Parameters to be passed to |
x |
A TableOne object created from |
These are very hacky functions. If used within an RMarkdown document,
KreateTableOne and KnitableTableOne should be called in a code chunk with
results='hide' to hide the plain test results printed from
tableone::CreateTableOne. The resulting data frame should be saved
as an object and used in a second code chunk for formatted printing.
Suggestions for improvement are welcomed.
The function is written to work with knitr::kable, but should be able
to work with other functions such as xtable::xtable.
Returns a data frame of character columns.
table1 = KreateTableOne(data=mtcars, strata='am', factorVars='vs') table1 knitr::kable(table1)table1 = KreateTableOne(data=mtcars, strata='am', factorVars='vs') table1 knitr::kable(table1)
num_compare gives details about the distribution of a numeric variable across subsets of the dataset
num_compare(y, grp, plot = c("density", "boxplot", "none"))num_compare(y, grp, plot = c("density", "boxplot", "none"))
y |
A numerical variable |
grp |
A categorical variable |
plot |
Type of plot to produce |
Returns a list including (1) group-wise summary statistics, (2) ANOVA decomposition, (3) eta-squared effect size, and (4) ggplot2 object, if requested.
v1 = rbinom(n=50, size=1, p=0.5) v2 = rnorm(50) num_compare(y=v2, grp=v1, plot='density')v1 = rbinom(n=50, size=1, p=0.5) v2 = rnorm(50) num_compare(y=v2, grp=v1, plot='density')
write_int formats a numeric input into an interval to be printed, e.g., in an RMarkdown document.
write_int(x, delim = "(", digits = 2)write_int(x, delim = "(", digits = 2)
x |
A length-2 numeric vector consisting of the endpoints of the interval or an n-row by 2-column matrix of endpoints. |
delim |
The bracket delimiters to surround the interval. Must be either a round bracket, square bracket, curly bracket, or angled bracket. |
digits |
Number of digits to round to (default to 2). Will keep trailing zeros. |
If a matrix is provided, the values in each row will be used to create a formatted interval.
Returns a character string of the form "(x[1], x[2])" (or supplied bracket delimiter).
write_int(x=c(1.2, 2.345)) write_int(x=c(1.2, 2.345), delim='[')write_int(x=c(1.2, 2.345)) write_int(x=c(1.2, 2.345), delim='[')
write_p formats a p-value for display in an RMarkdown document.
write_p(x, digits = 2)write_p(x, digits = 2)
x |
A length-1 numeric or a list-like object with element named |
digits |
Number of digits to round to (default to 2) |
If x < 10^(-digits), then the result is the string p < 10^(-digits) in decimal notation.
Returns a LaTeX-formatted string to report a p-value to the specified number of digits.
write_p(0.2345) write_p(0.000234) x = rnorm(10) test1 = t.test(x) write_p(test1)write_p(0.2345) write_p(0.000234) x = rnorm(10) test1 = t.test(x) write_p(test1)