Fix #475 - add flowchart #545
Conversation
|
@jwildfire - added a possible implementation but looking for any additional specs you can provide for what you'd like to see here. |
Merge branch 'dev' of github.com:Gilead-BioStats/gsm into fix-475 # Please enter a commit message to explain why this merge is necessary, # especially if it merges an updated upstream into a topic branch. # # Lines starting with '#' will be ignored, and an empty message aborts # the commit.
Merge branch 'dev' of github.com:Gilead-BioStats/gsm into fix-475 # Please enter a commit message to explain why this merge is necessary, # especially if it merges an updated upstream into a topic branch. # # Lines starting with '#' will be ignored, and an empty message aborts # the commit.
Merge branch 'dev' of github.com:Gilead-BioStats/gsm into fix-475 # Please enter a commit message to explain why this merge is necessary, # especially if it merges an updated upstream into a topic branch. # # Lines starting with '#' will be ignored, and an empty message aborts # the commit.
|
It's running for me, and it looks really nice! It would be a lot more intuitive if we had consistent verbiage. For instance, |
|
@jwildfire - the code to create a single flowchart is pretty nested inside library(DiagrammeR)
lAssessments <- MakeAssessmentList()
lData <- list(
dfSUBJ = clindata::rawplus_subj,
dfAE = clindata::rawplus_ae,
dfPD = clindata::rawplus_pd,
dfCONSENT = clindata::rawplus_consent,
dfIE = clindata::rawplus_ie
)
lTags <- list(
Study = "myStudy"
)
lMapping <- clindata::mapping_rawplus
ae_assessment <- RunAssessment(lAssessments$ae, lData = lData, lMapping = lMapping, lTags = lTags)
ae_assessment$lResults$flowchart
|
|
It's running for me, too, but I have to |
|
Running fine in RStudio - sorry for confusion! Will continue to a full review soon. |
R/util-RunAssessment.R
Outdated
| node_df <- bind_rows(node_df, | ||
| map_df(step$inputs, ~tibble( | ||
| assessment = lAssessment$name, | ||
| n_step = stepCount, | ||
| name = step$name, | ||
| inputs = .x, | ||
| n_row = nrow(lData[[.x]]), | ||
| n_col = ncol(lData[[.x]]))) %>% | ||
| mutate(checks = result$lChecks$status, | ||
| from = stepCount, | ||
| to = length(lAssessment$workflow[[stepCount]]$inputs) + stepCount, | ||
| n_row_end = ifelse(!is.null(result$newRows), result$newRows, NA)) | ||
| ) | ||
|
|
There was a problem hiding this comment.
I think I'd vote to move this to Visualize_Workflow if possible ...
R/util-RunAssessment.R
Outdated
| lAssessment$lResults$flowchart <- Visualize_Workflow( | ||
| lAssessment = lAssessment, | ||
| lResult = result, | ||
| dfNode = node_df | ||
| ) |
There was a problem hiding this comment.
result is already embedded in lAssessment, no? Seems like we can just pass lAssessment and then have Visualize_Workflow calculate node_df
R/util-FilterDomain.R
Outdated
| if (bReturnChecks) { | ||
| return(list(df = df, lChecks = checks)) | ||
| if(!exists("newRows")) newRows <- NA |
There was a problem hiding this comment.
A little bit confused about the need for this ... can we not just use nrow() on the dataset that is returned.
R/util-RunAssessment.R
Outdated
| stepCount <- stepCount + 1 | ||
| } | ||
| } else { | ||
| cli::cli_alert_warning("Workflow not found for {lAssessment$name} assessment - Skipping remaining steps") | ||
| lAssessment$bStatus <- FALSE | ||
| } | ||
|
|
||
| if(lAssessment$bStatus) { | ||
| lAssessment$lResults$flowchart <- Visualize_Workflow( |
There was a problem hiding this comment.
I don't have a super strong opinion, but in my mind the flowchart fits better under lChecks than lResults. It's really more of a debugging/workflow tool than a part of the core KRI workflow.
There was a problem hiding this comment.
Or better yet attached to workflow:
Also should checks be in lChecks?
|
Add to |
|
@jwildfire - I think this is as far as it makes sense to take things for now. Feel free to review again, wait to discuss further, or merge if it looks good. To run with complete data: study <- Study_Assess()
study$ae$lChecks$flowchartError for subject-level data: lData <- list(
dfSUBJ = clindata::rawplus_subj %>% select(-SubjectID),
dfAE = clindata::rawplus_ae
)
x <- Study_Assess(lData = lData)
x$ae$lChecks$flowchartError for domain-level/missing data:
lAssessments <- MakeAssessmentList()
lData <- list(
dfSUBJ = clindata::rawplus_subj,
dfAE = clindata::rawplus_ae %>% select(1)
)
lTags <- list(
Study = "myStudy"
)
lMapping <- clindata::mapping_rawplus
sae_assessment <- RunAssessment(lAssessments$sae, lData = lData, lMapping = lMapping, lTags = lTags)
sae_assessment$lChecks$flowchart |
|
More code to test if helpful: dfAE <- clindata::rawplus_ae
dfCONSENT <- clindata::rawplus_consent
dfIE <- clindata::rawplus_ie
dfPD <- clindata::rawplus_pd
dfSUBJ <- clindata::rawplus_subj
# AE missing SubjectID column
dfAE$SubjectID <- NULL
# Consent column CONSENT_VALUE has NA
dfCONSENT$CONSENT_VALUE[1:10] <- NA
# IE column IE_VALUE missing
dfIE$IE_VALUE <- NULL
lData <- list(
dfAE = dfAE,
dfCONSENT = dfCONSENT,
dfIE = dfIE,
dfPD = dfPD,
dfSUBJ = dfSUBJ
)
x <- Study_Assess(lData = lData)
x$ae$lChecks$flowchart
x$consent$lChecks$flowchart
x$ie$lChecks$flowchart |
Overview
Initial bare bones implementation for flowchart
Test Notes/Sample Code