2016-07-08

[knighttime]

@andkov

Friday July 8th, 9:30am

Meeting Agenda

I will update my MAP files before the meeting.

1. Go through the files 0-7 : address: any changes or errors, clean ups
2. Some help with files 5-7
3. Discuss plans on how to merge the new variables into MAP curator github
4. Wait for new data pull, then work on bringing in the new data into MAP curator

[andkov]

NOTE:

• you must filter out the observation form different studies (non MAP) before you do the merger between the wide and the long file. Otherwise you will produce duplicated rows that would not to be removed later

[andkov]

Amelia - a package to visualize missing patterns http://www.inside-r.org/packages/cran/Amelia/docs/missmap

#check missing data
#this can be intensive, save everything in case computer crashes 
Amelia::missmap(dsw, main = "Missing values vs observed") 

[andkov]

tighten up the subsetting and creation of new variables:

# ---- tweak-data ---------------------------------------
d <- ds %>% 
  dplyr::mutate(
    vital_status    =  ifelse(is.na(age_death) , 0 , 1), # jamie, this is it.
    dementia_status =  ifelse(is.na(age_death) , 0 , 1),
    stroke_status   =  stroke_cum,
    path_status     =  ad_reagan,
    apoe_genotype   =  ifelse(apoe_genotype %in% c(44,34,24), 1, 0),
    group_smell     =  ordered(cut(total_smell_test, c(0,5,10,12), 
                                   labels=c("anosmic", "hyposmic", "normosmic")))
)

[andkov]

x: age y: mmse facets: smell grouping and apoe status color : cumulative lifetime stroke status

[andkov]

x: age y: mmse facets: smell group and pathology status (1 = most pathologies, 3 = least ) color: apoe status

[knighttime]

These graphs are super! is there any way to get rid of the NA column? - will pick this up next friday

[andkov]

Yes, just modify the dplyr::filter() call

[ampiccinin]

re: filtering non-MAP studies - could this also be resolved by retaining all data from all studies (which someone might like to do)?

[knighttime]

@andkov - got it!!! Thanks + I bought that graphing book 👍

[knighttime]

@ampiccinin We discovered that the duplicates were mostly in MARS. So subsetting down to just MAP should fix the duplicates - though a bit more experimentation is needed to figure out why both Cassandra and Rebecca found duplicates in theirs - I thought that their data sets were already subsetted down to just MAP - so we need to check that.