Skip to contents

This function provide a report showing all outlier values for each numerical fields. The function will try to automatically determine the type of distribution (between Normal and Log-Normal) based on the difference between mean and median between untransformed normalized and log transformed normalized distribution.

Usage

surveyOutliers(
  ds = NULL,
  enumeratorID = NULL,
  sdval = 2,
  reportingColumns = c(enumeratorID, uniquerespondantID),
  enumeratorCheck = FALSE
)

Arguments

ds

dataset containing the survey (from kobo): labelled data.frame

enumeratorID

name of the field where the enumerator ID is stored: string

sdval

(Optional, by default set to 2) number of standard deviation for which the data within is considered as acceptable: integer

reportingColumns

(Optional, by default it is built from the enumeratorID and the uniquerespondantID) name of the columns from the dataset you want in the result: list of string (c('col1','col2',...))

enumeratorCheck

(Optional, by default set to FALSE) specify if the report has to be displayed for each enumerator or not: boolean (TRUE/FALSE)

checkperiod

if not null number of day before today when the check should be made

surveyConsent

name of the field in the dataset where the survey consent is stored: string

consentForValidSurvey

value defined in the kobo form to acknowledge the surveyed person gave his consent: string

uniquerespondantID

name of the field where the survey unique ID is stored: string

Value

dst same dataset as the inputed one but with survey marked for deletion if errors are found and delete=TRUE (or NULL)

ret_log list of the errors found (or NULL)

var a list of value (or NULL)

graph graphical representation of the results (or NULL)