Open KatrinCoboeken opened 5 years ago
The purpose of the quantile-quantile plot is to determine whether the residuals are drawn from a normal distribution.
Residuals are calculated as linear or logarithmic difference of observed - predicted
Data
· Data.frame/data.table with following columns
o predicted (double) value of the predicted output to show in the unit of display, so no unit conversion is needed
o observed (double) value of the predicted output to show in the unit of display, so no unit conversion is needed
o Group (ordered factor) (character)
it is used to group individual results,
it is also used for legend (e.g. “Pediatric population”,”Adult Population”)
order of factor is used for legend order
· simulationResult.Metadata metada of Simulation Result
List with following entries:
o Value_unit: character
FigureConfiguration:
· value_label: character (e.g. ‘plasma concentrations’)
· scaleToCalculateDifference: character (default ‘lin’)
· colorscale_groups (default NULL, then take R defaults) defines colors for different groups
· shape_vector (default NULL, then take default) defines shape of residuals per identifier
List with entries
· Figure: Figure object
If scaleToCalculateDifference = ‘lin’
residuals = observed – prediction
If scaleToCalculateDifference = ‘log’
residuals = log(observed) – log(prediction)
· plot quantiles of residuals vs quantiles of a normal distribution
· Plot reference line
· Set color by group
· Set shape by group
· group name
· according to Xlabel and Ylabel issue #27 Figure: plot time profile
· Ylabel: ‘perctentiles residuals of ’ +
‘(observed – predicted)’ or ‘(log(observed) –log( predicted))’
Example
ju-rgen
commented
5 years ago 
ju-rgen
commented
5 years ago
Figure: quantile-quantile plot
The purpose of the quantile-quantile plot is to determine whether the residuals are drawn from a normal distribution.
Residuals are calculated as linear or logarithmic difference of observed - predicted
Arguments:
Data
· Data.frame/data.table with following columns
o predicted (double) value of the predicted output to show in the unit of display, so no unit conversion is needed
o observed (double) value of the predicted output to show in the unit of display, so no unit conversion is needed
o Group (ordered factor) (character)
it is used to group individual results,
it is also used for legend (e.g. “Pediatric population”,”Adult Population”)
order of factor is used for legend order
· simulationResult.Metadata metada of Simulation Result
List with following entries:
o Value_unit: character
FigureConfiguration:
· value_label: character (e.g. ‘plasma concentrations’)
· scaleToCalculateDifference: character (default ‘lin’)
· colorscale_groups (default NULL, then take R defaults) defines colors for different groups
· shape_vector (default NULL, then take default) defines shape of residuals per identifier
Return:
List with entries
· Figure: Figure object
Details:
Calcualtion of residuals
If scaleToCalculateDifference = ‘lin’
residuals = observed – prediction
If scaleToCalculateDifference = ‘log’
residuals = log(observed) – log(prediction)
Plotting of simulated results
· plot quantiles of residuals vs quantiles of a normal distribution
· Plot reference line
· Set color by group
· Set shape by group
Legend should contain
· group name
Xlabel and Ylabel
· according to Xlabel and Ylabel issue #27 Figure: plot time profile
· Ylabel: ‘perctentiles residuals of’ +
‘(observed – predicted)’ or ‘(log(observed) –log( predicted))’