cleaning.R

In gavandrewj/pdphApp: pdphApp

library(readxl)
library(tidyverse)

dataset <- read_excel('C:/Users/gavin/Desktop/Projects/shonette/dataset.xlsx')

originalnames <- names(dataset)

#convert the names to lowercase
names(dataset) <- tolower(names(dataset))



#convert the factor types
dataset$race <- factor(dataset$race,
                       levels = c(1,2,3,4),
                       labels = c('African',
                                  'Amerindian',
                                  'East Indian',
                                  'Mixed/Other'))                    

dataset$`type of needle` <- factor(dataset$`type of needle`,
                                   levels = c(0,1,2),
                                   labels = c('Not mentioned',
                                              'Quinckie',
                                              'Pencil'))

dataset$guage <- factor(dataset$guage)


dataset$provider <- factor(dataset$provider,
                           levels = c(1,2,3,0),
                           labels = c('Doctor',
                                      'Nurse',
                                      'Doctor and Nurse',
                                      'Not mentioned'))


dataset$`fronto-occipital` <- factor(dataset$`fronto-occipital`,
                                     levels = c(1,0),
                                     labels = c('Yes',
                                                'No'))


dataset$positional <- factor(dataset$positional,
                             levels = c(1,0),
                             labels = c('Yes',
                                        'No'))

dataset$`neck pain` <- factor(dataset$`neck pain`,
                              levels = c(1,0),
                              labels = c('Yes',
                                         'No'))


dataset$fluids <- factor(dataset$fluids,
                         levels = c(1,0),
                         labels = c('Yes',
                                    'No'))


dataset$nsaid <- factor(dataset$nsaid,
                        levels = c(1,0),
                        labels = c('Yes',
                                   'No'))


dataset$caffiene <- factor(dataset$caffiene,,
                           levels = c(1,0),
                           labels = c('Yes',
                                      'No'))


dataset$opioid <- factor(dataset$opioid,
                         levels = c(1,0),
                         labels = c('Yes',
                                    'No'))


dataset$`blood patch` <- factor(dataset$`blood patch`,
                                levels = c(1,0),
                                labels = c('Yes',
                                           'No'))

dataset$resolution <- factor(dataset$resolution,
                             levels = c(0,2,3,4),
                             labels = c('No resolution',
                                        '2 Days',
                                        '3 Days',
                                        '4 Days'))


dataset$`headache onset` <- factor(dataset$`headache onset`,
                                   levels = c(0,1,2),
                                   labels = c('No Headache','First Day','Second Day'))


dataset$guage <- factor(dataset$guage,
                        levels = c(0,22,23,24,25,27),
                        labels = c('Not mentioned','22','23','24','25','27'))





names(dataset) <- originalnames


tolabel <-   names(dataset)
tolabel[1] <- 'Patient ID' 
arsenal::labels(dataset)  <- tolabel

tonames <- c( "Pt.Id"         ,   "Age"         ,      "Race"   ,           "TypeofNeedle",   
"Guage"         ,    "Attempt"     ,      "Provider"      ,    "HeadacheOnset",   
"FrontoOccipital"  ,"Positional"   ,     "NeckPain"  ,       "Fluids"  ,         
"Nsaid"          ,   "Caffiene"       ,   "Opioid"         ,   "Resolution",       
 "BloodPatch"   ,    "DischargeSummary")

tonames[1] <- 'PT.ID'
tonames[4] <- 'TypeofNeedle'
tonames[8] <- 'HeadacheOnset'
tonames[17] <- 'BloodPatch'
tonames[18] <- 'DischargeSummary'

names(dataset) <- tonames

saveRDS(dataset,'C:/Users/gavin/Documents/GitHub/pdphApp/dataset.Rds')