tidal-modelling / tidal Goto Github PK

Describe the bug
A clear and concise description of what the bug is.

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. mac, windows, linux]
• R Studio, R package, Online version, Desktop app,
• Browser (if applicable) [e.g. chrome, safari]

Additional context
Add any other context about the problem here.

add CIs to data exploration page

Describe the bug
A clear and concise description of what the bug is.

To Reproduce
Steps to reproduce the behavior:

1. Data exploration page
2. Add a continuous covariate

** Error message **
Error in mutate: ℹ In argument: er_total_t1 = (function (length = 0L) ....

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. mac, windows, linux]
• R Studio, R package, Online version, Desktop app,
• Browser (if applicable) [e.g. chrome, safari]

Additional context
Add any other context about the problem here.

On the data exploration page the user has the option to Z-score standardise any continuous variables.

If users have column names containing spaces these are replaced with underscores. If column names contain "(", ")" or "*" these are replaced with empty strings "". Could you add this somewhere to the documentation please?

Include interaction terms to estimate the plot line

Not urgent, but I've figured a way to edit the global default ggplot theme here:
https://github.com/AmeliaES/TIDAL/blob/3a3a00a124f7721bd2499f5c98b175aba3e9348a/inst/TIDALapp/app.R#L12-L14
I'm not sure the plots look particularly pretty... if you're bored feel free to have a look if you can make them look nicer :-)

If data has cells with empty strings ie "" there is no option to select that column in the data exploration page. But only when checking if the column is numeric or not.

Tried to add survey weights as an option in lmer, but variables not explicitly available in the global environment can't be parsed to the weights argument in lmer (see below for details). Please see branch F_lmerWeights. To install from this branch specifically you can do devtools::install_github("AmeliaES/TIDAL", ref = "F_lmerWeights")

I have found what the problem is and show a reproducible example outside the Shiny environment using the "MCS_Wide_6_Weights" dataset Alex sent me below:

library(lme4)
library(dplyr)

newModelData <- read.csv("MCS_Wide_6_Weights_LongFormat_2023-08-22.csv")
formCodeCovars <- "score ~ age + ( 1 + age | subject )"

testlmer <- function(weightsTest){

  weightsVec <- pull(newModelData, !!sym(weightsTest))

  fit <- lmer(formula = formCodeCovars,
       REML=FALSE ,
       data = newModelData,
       control=lmerControl(optimizer="bobyqa",
                           optCtrl=list(maxfun=2e5)),
       weights =  weightsVec)
  return(fit)
}

testlmer(weightsTest = "WEIGHT2")

testlmer <- function(weightsTest){

  fit <- lmer(formula = formCodeCovars,
              REML=FALSE ,
              data = newModelData,
              control=lmerControl(optimizer="bobyqa",
                                  optCtrl=list(maxfun=2e5)),
              weights =  newModelData[[weightsTest]])
  return(fit)
}

testlmer(weightsTest = "WEIGHT2")

# -------------------------------------
# The following works:

weightsTest <- "WEIGHT2"
lmer(formula = formCodeCovars,
              REML=FALSE ,
              data = newModelData,
              control=lmerControl(optimizer="bobyqa",
                                  optCtrl=list(maxfun=2e5)),
              weights =  newModelData[[weightsTest]])


# -------------------------------------
# Conclusion:
# lmer can only pass variables in the global environment to the weights argument
# if the variable is in eg. a function or shiny environment then it can't find it.

# the only thing that did work in the shiny environment was explicitly writing the column variable name:
lmer(formula = formCodeCovars,
     REML=FALSE ,
     data = newModelData,
     control=lmerControl(optimizer="bobyqa",
                         optCtrl=list(maxfun=2e5)),
     weights =  WEIGHT2)

Allow users to upload SPSS and STATA files, .sav and .dta

add violin plots for variables users select for them to get a visual description of their data, check the range looks sensible, any outliers etc.

Describe the bug
Error if spaces in column name variables

Replace spaces with underscores after initial upload of data frames. Check for both wide and long upload options.

Describe the bug
A clear and concise description of what the bug is.

AUC and scores at ages doesn't work if interaction variable contains NA

To Reproduce
Steps to reproduce the behavior:

1. Upload data with missing values in the categorical Sex column.
2. Run a linear model as standard
3. Go to the Interaction Variable page, select Sex as categorical variable.
4. Interact as usual with the Scores at ages and AUC tabs.
5. Note how the tables dont show or the difference in males and females for AUC. Note how there is also an NA plotted in the legend but no plot line for NA.

Expected behavior
Expect to see the table for scores at ages, and 0,1 to be automatically selected in AUC with a sentence stating the difference in AUC between those 2 levels.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• R package

Additional context
Add any other context about the problem here.

Add interpretation of parameters. E.g. for each increase in X (enter the variable they put in), Y (enter the outcome variable) increased by.... accounting for... etc.

accessibility feature

If a .xlsx file is uploaded then the app crashes. Include a trycatch statement to prevent this crash.

Describe the bug
data exploration plot not updated when changing model type and clicking on run model.

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. mac, windows, linux]
• R Studio, R package, Online version, Desktop app,
• Browser (if applicable) [e.g. chrome, safari]

Additional context
Add any other context about the problem here.

allow user to edit ggplots, eg. colours, labels of axes, labels in legend.

download report to include differences in scores at ages and AUC now we have included CIs. Also noticed plot isn't rendering for scores at ages.

check it works even when users don't interact with scores at ages and/or AUC tabs

If you click on Download Report button in either the data exploration page or the interaction variable page but dont interact with the scores at ages and AUC tabs then the Rmd wont render.

Make a way that if these two plots and tables aren't produced the Rmd will still render without these variables.

Working on F_DownloadResults branch.

makes plots more legible by increasing text size and axis labels

From creating Window's installer it was found that the Download Report feature didn't work, even when they tried using RStudio on Windows. See below:
"some issues with Rstudio on windows and producing PDFs from the download data button. The root cause appears to be two things are not happening:
1 – Tidal.png is not being copied t the build directory
2 – There’s a call to \includegraphics a little after “number-of-observations-and-groups”. It calls for a pdf and folder to be copied. These are never created on windows. From R studio"

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

unsure how to add 95% CIs to this, or if we need to @kwongsiufung might need some help on this one please, not sure glht will work? or is it a different expression we give to this function?

Describe the bug
Spaces and maybe other characters in the level names of a categorical variable causes an issue in plotting on interaction variable page.

lincom or https://www.rdocumentation.org/packages/multcomp/versions/1.4-23/topics/glht

User reported installing the app onto a VM but only being able to access local files when they need to access the files on the VM:

"I was able to install the package on our cloud where we have all our data. I would like to analyse data on the cloud using the app, but when I select the option to browse folders to select data, it directs me to the local folders even though the app is installed on the cloud.

It's an OpenStack cloud and we can't access the cloud files locally."

Having a quick look at this issue I found this package that might offer a solution: https://github.com/thomasp85/shinyFiles

Something to indicate the code is running when users click on the button eg https://community.rstudio.com/t/progress-indicator-for-event-reactive-without-loops/2343/2

tidal logo too big on download report
change figure size in .Rmd

By default the reference level is whatever is first when factorising it. Change this so the user can specify what level they want as the reference variable in their model.

Implement this by:
"relevel(factor, ref = "reference level") - it's from the stats package"