kyle.belanger/DHSC-Capstone
1
0
Fork 0
Code Issues 1 Pull requests Projects Releases Packages Wiki Activity Actions

Update 0-data_prep.R

Browse source
This commit is contained in:
Kyle Belanger 2023-01-06 15:48:47 -05:00
parent 8a348759cc
commit de69edfecc
1 changed files with 17 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
ML/0-data_prep.R
View file

@ -60,6 +60,17 @@ test_list_bmp <- c(
# load data ---------------------------------------------------------------
# load patients first to add to lab values
patients <- dplyr$tbl(db, "patients") %>%
dplyr$select(-anchor_year, -anchor_year_group, -dod) %>%
dplyr$collect()
#this function is failing if run as part of the DB query
# Recoding Male = 1, Female = 2
patients <- patients %>%
dplyr$mutate(dplyr$across(gender, ~dplyr$recode(gender, "M" = 1, "F" = 2)))
# most likely will not use this as there are not as many complete rows. However
# gathering it just in case.
# first is using specimen id, usable data set is using chart time as it appears
@ -90,7 +101,8 @@ ds_cmp <- dplyr$tbl(db, "labevents") %>%
#this keeps failing if run as part of the above query. Moving here to keep going
# keeps only rows that have values for all columns
ds_cmp <- ds_cmp %>% dplyr$filter(dplyr$if_all(.fns = ~!is.na(.)))
ds_cmp <- ds_cmp %>% dplyr$filter(dplyr$if_all(.fns = ~!is.na(.))) %>%
dplyr$left_join(patients, by = c("subject_id" = "subject_id"))
ds_bmp <- dplyr$tbl(db, "labevents") %>%
@ -107,6 +119,10 @@ ds_bmp <- dplyr$tbl(db, "labevents") %>%
ds_bmp <- ds_bmp %>% dplyr$filter(dplyr$if_all(.fns = ~!is.na(.)))
# close database ----------------------------------------------------------
dbDisconnect(db)