(Link back to Version 4 Archive)
New parameters (i.e. arguments) added to the function: ds.glm
: data
, weights
and offset
, checks
, startBetas
.
data
allows for user to specify the name of an optional data frame that holds the variables in the regression formula so one can can write for example:
ds.glm(formula='DIS_DIAB~GENDER', data='D', family='binomial') |
which is equivalent to
ds.glm(formula='D$DIS_DIAB~D$GENDER', family='binomial') |
In the previous release only the latter command was possible when the variables in the regression formula were held in a data frame 'D'.
weights
to specify the name of a numeric variable of 'prior weights' to be used in the model fitting process.offset
to specify the name of a known component to be included in the linear predictor.checks
to help with error tracking. These checks are switched off by default. If set to TRUE thorough checks are carried out before the process starts.startBetas
to specify starting values for the parameters in the linear predictor. In earlier versions this parameter was named startCoeff
.The Functions were re-written from scratch to implement a new way of masking 'invalid' cells whilst reporting correct total counts (i.e. to address the issues raised in BioSHaRe annual meeting, Athens 2014).
The new versions of these two functions return a more tidy output with improved error reporting and improved flexibility.
In the previous version of this function a t.test was only possible for two continuous vectors.
The new version allows for the comparison of the mean values of a continuous vector across the categories of factor as in the below example where both vectors are in a data frame 'D':
# The continuous vector is on the left of the formula whilst the factor is on the right side ds.tTest(x='D$PM_BMICONTINUOUS~D$GENDER') |
New parameter (i.e. argument) added to the function: completeCases
:
completeCases
if set to TRUE allows to construct a complete table (i.e. a data frame where rows with missing values are excluded)In the new version of this function it is possible to compute the mean and standard deviation of a continuous vector across the categories of a factor as follows:
# The continuous vector is on the left of the formula whilst the factor is on the right side. In the below example both vector are in a data frame named 'D'. ds.meanByClass(x='D$LAB_HDL~D$GENDER') |
This addition was implemented for improved flexibility and is equivalent to the below syntax which still works and should be used to specify more than one outcome or covariate.
ds.meanByClass(x='D', outvar=c('LAB_HDL'), covar=c('GENDER')) |
This function replaces the previous ds.subclass
Returns the number of missing values in a given vector.
To replaces missing values in a given vector by specific value(s).
To display the names of the objects stored/defined on the server side.
This function has been replaced by ds.subsetByClass
ds.subclass
replaced by ds.subsetByClass
). Please amend your previous scripts/commands to adopt the new functions names and syntax to avoid errors.Remarks:
|
If you are the administrator of the opal server for a cohort: Please deploy the new package in your servers. Install all the packages via opal as follows:
Administration
on top-right cornerDataSHIELD
Add Package
Install
Action
, click on Publish methods
A /wiki/spaces/DSDEV/pages/12943456 is available.