change sigmoid point in assess_dependencies

[Edward-Gillian]

Description

1. Change dependencies sigmoid point to 20 in the function assess_dependencies (https://github.com/pharmaR/riskmetric/blob/master/R/assess_dependencies.R#L87)

2. update the examples for this function in lines 73, 76, 79

3. Regenerate the documentation

Use Case

This will allow for a more realistic number of dependent packages as five can be considered as too low.

[dgkf]

I'm not personally opposed to changing metrics to make them more useful, but I do think we'd need to be extra cautious with communication so that users are very mindful of changes to the metric calculations.

If we were to change a metric, I'd prefer to have some empirical backing for the decision. One starting point might be to use riskscore to analytically derive a reasonable midpoint.

Aside from having the right "feel", would you be able to motivate this change with some analytic rationale?

[Edward-Gillian]

I have some evidence that the sigmoid point should be adjusted upwards.

We have riskmetric data on some internal packages and the following packages:

pkg_name	pkg_version
checkmate	2.3.1
covr	3.6.0
diffdf	1.0.4
DT	0.32
ggiraph	0.7.10
ggplot2	3.4.4
glue	1.7.0
here	1.0.1
httr2	1.0.0
knitr	1.45
lubridate	1.9.2
magrittr	2.0.3
MultiGroupSequential	1.1.0
pandoc	0.2.0
pkglite	0.2.1
purrr	1.0.2
utf8	1.2.3
rhino	1.5.0
riskmetric	0.2.1
stringr	1.5.1
styler	1.10.2
testthat	3.2.1
tibble	3.2.1
vctrs	0.6.5
waldo	0.5.2
xlsx	0.6.5
Tplyr	1.1.0
admiral	1.0.0
rtables	0.6.6
dplyr	1.1.0
circlize	0.4.13
ComplexHeatmap	2.9.4
reda	0.5.4
emmeans	1.10.1.90001
geeasy	0.1.1
MASS	7.3-60.0.1
mgcv	1.9-1
nlme	3.1-164
patchwork	1.2.0
PWEALL	1.3.0.1
RBesT	1.7-3
brms	2.21.0
rjags	15-Apr
tmle	2.0.0
broom	1.0.5
ggsurvfit	1.0.0
survival	3.5-8
flextable	0.8.1
forcats	1.0.0
huxtable	5.5.6
survminer	0.4.9
ggpmisc	0.5.2
mmrm	0.3.11
rpact	3.5.1
farver	2.1.1
marginaleffects	0.16.0
rms	6.7-1
arm	1.13-1
ashr	2.2-63
brglm2	0.9.1
xCell	1.1.0
BOIN	2.7.2
clubSandwich	0.5.10
psborrow2	0.0.3.2
reshape2	1.4.4
censReg	0.5-36
escalation	0.1.8
geepack	1.3.9
ggfortify	0.4.15
quantreg	5.97
cubature	2.0.3
DEoptim	2.2-7
ggalluvial	0.12.2
gtable	0.3.4
R2jags	0.7-1
simsurv	1.0.0
ggbreak	0.1.1
numDeriv	2016.8-1.1
PBIR	0.1-0
survRM2	1.0-4
lme4	1.1-15
doParallel	1.0.17
foreach	1.5.2
gridExtra	2.3
bayesplot	1.10.0
coda	0.19-4
future	1.33.1
gridGraphics	0.4-0
gridtext	0.1.4
mice	3.15.0
Rcpp	1.0.11
psbcGroup	1.5
mvnfast	0.2.7
psbcGroup	1.5
RcppEigen	0.3.3.9.4
splines2	0.5.1

When calculating the average number of dependencies looking at Imports and LinkingTo , the average number of dependencies is 8.066038.

When calculating the average number of dependencies looking at Imports , LinkingTo , and Suggests, the average number of dependencies is 18.28302.