Something wrong when using goodness_of_fit

Basic Usage
1

Hey All,

I am trying to analyze my fitted data using goodness_of_fit, but it does not seem to work with the dataframe I made. I get the error: goodness_of_fit cannot be plotted into a single axis.` I have a dataframe being input, so I don’t understand why it says there is only a single axis.

Here is my code:

## Fitting weibull model to simulated population
@time fit_Weibull = [fit(model_Weibull, pop_Weibull[i], param_Weibull) for i in 1:length(pop_Weibull)]

## Post processing
inspect_Weibull = reduce(vcat, DataFrame(inspect(id)) for id in fit_Weibull)
GoF_Weibull = reduce(vcat, DataFrame(goodness_of_fit(inspect_Weibull)))
2

you cannot post-process a DataFrame. It has to be of Type FittedPumasInspection. Try this (untested)

## Fitting weibull model to simulated population
@time fit_Weibull = [fit(model_Weibull, pop_Weibull[i], param_Weibull) for i in 1:length(pop_Weibull)]

## Post processing
inspect_Weibull = map(ft -> inspect(ft), fit_Weibull)
GoF_Weibull = map(insp -> goodness_of_fit(insp), inspect_Weibull) # will result in a vector of GOF

you can index in and check each GOF like this GoF_Weibull[1], GoF_Weibull[2] ...
if each of them happen to be just the figure, then you can try

report(GoF_Weibull, clean = false)
3

Thank you. This makes sense. The inspect lines works but I am getting a “Makie.convert_arguments” error in the goodness_of_fit line.

4

can you just try observations_vs_predictions instead of goodness_of_fit

5

Thanks for the suggestion. It still gives the same error though.
Here is the full code:

# Practice fitting Weibull model using Pumas
using Random, Pumas, PumasUtilities, CairoMakie, AlgebraOfGraphics, DataFramesMeta, DataFrames, CSV

## Read in simulated population of 6 formulations with fdiss (percent drug dissolved)
sim_Emax = CSV.read("/home/jrun/data/code/Practice/ExercisePK44_1/Data/Emaxpop.csv", DataFrame)

## Create pupulation
pop_Weibull = read_pumas(sim_Emax, observations=[:dv_fdiss], covariates=[:dtd50], event_data=false)

## Model definition Weibull
model_Weibull = @model begin
    @metadata begin
        desc = "Dissolution model Weibul"
        timeu = u"minute"
    end

    @param begin
        "Maximum percent drug dissolved (%)"
        θfdissmax ∈ RealDomain(lower=0)
        "Time required for 50 percent of the drug to dissolve (minutes)"
            θtd50 ∈ RealDomain(lower=0)
        "Additive RUV"
            σadd ∈ RealDomain(lower=0)
        "sigmoidicity factor"
               θB ∈ RealDomain(lower=0)
    end

    @covariates dtd50

    @pre begin
         fdissmax = θfdissmax
             td50 = θtd50 * (1 + dtd50)
                B = θB
    end

    @derived begin
        fdiss = @. fdissmax * (1 - exp((t/td50)*B))
        """
        Percent dissolved in vitro
        """
        dv_fdiss ~ @. Normal(fdiss, σadd)
        
    end
end

## Initial parameters
param_Weibull = (θfdissmax = 100.0, θtd50 = 10, σadd = 0.01, θB = 0.5)

## Fit Weibull model to simulated population from Emax model
@time fit_Weibull = [fit(model_Weibull, pop_Weibull[i], param_Weibull) for i in 1:length(pop_Weibull)]

## Post processing
inspect_Weibull = map(ft -> inspect(ft), fit_Weibull)
GoF_Weibull = map(insp -> observations_vs_predictions(insp), inspect_Weibull)

## Plot 
@chain inspect_Weibull begin
    @rsubset :time ∈ [1,5,10,15,30,60,90,120]
    data(_) *
    mapping(:time => "Time(minutes)",
        :dv_fdiss => "Percent dissolved",
        color = :dtd50 => nonnumeric => "TD50") *
    visual(ScatterLines, linewidth = 4, markersize = 12)
    draw(axis = (;xticks = 0:15:120, yticks = 0:10:100))
end

## Output
report((fit_Weibull[1], infer_Weibull[1], inspected_Weibull[1]), output="/home/jrun/data/code/Practice/ExercisePK44_1/Reports/Emaxgamma_report")
6

can you post the full error please? The entire error that comes in the REPL

7

ERROR: Makie.convert_arguments for the plot type Lines{Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}, Float64}} and its conversion trait PointBased() was unsuccessful.

The signature that could not be converted was:
::Vector{Float32}, ::Float64

Makie needs to convert all plot input arguments to types that can be consumed by the backends (typically Arrays with Float32 elements).
You can define a method for Makie.convert_arguments (a type recipe) for these types or their supertypes to make this set of arguments convertible (See Plot Recipes).

Alternatively, you can define Makie.convert_single_argument for single arguments which have types that are unknown to Makie but which can be converted to known types and fed back to the conversion pipeline.

8

The full error is much longer. let me know if you need that, though the REPL does not load the whole error message.

9

I would try two things: 1) hide the fdiss variable from output using := 2) properly specify weibull function. Change this:

to:

fdiss := @. fdissmax * (1 - exp(-(t/td50)^B))
10

Thank you for spotting that mistake.

11

@KVTobin can you provide that complete error message, thanks. There’s definitely something odd happening with the argument types internally, so a more complete error will help me better narrow down the issue.

12

It seems to be working now, but is giving the output as a vector of {Makie.FigureAxisPlot}. How do I visualize them?

13

You should be able to index into that vector to plot each directly in the vscode plot pane.

14

Understood. Thank you for all your help!