0

I have a survey data which has 18.406 rows and 247 columns. It has too many likert scale questions. And of course it has NAs.

I checked similar questions and I saw an answer in that topic: Cluster Analysis in R with missing data But it doesn't work for me.

I am trying to use Fuzzy clustering method. As a first step, I subset data with "select" function for the questions I want to examine. Now my data has "integer" values. And after that I tried to have Dissimilarity Matrix with "daisy" function. This allows NAs. After this step, I used "fanny". The first argument needs Dissimilarity Matrix -or data frame- and in "fanny" , NAs are not allowed. Although removing rows with NAs is not the solution I'm looking for.

What am I doing wrong?

Cheers,

www
  • 38,575
  • 12
  • 48
  • 84
user3519155
  • 1
  • 1
  • 1
    It may help to think about what NA values mean in the context of your survey items. Are they simply skipped questions; are they where the respondent doesn't know how to answer; are they missing because they are not applicable; etc. Once you have the answer to that, think about what the NA means in relation to the values on your Likert scales. If, for example, You are asking the respondent to rate satisfaction with a product on a 1-5 scale, and the respondent hasn't used the product (i.e., NA), how would you expect to contrast them with someone who rates the product a 1 or a 5? – Amw 5G Aug 10 '15 at 11:39
  • hello @Amw5G, thank you for your answer. It's been a while I didn't write here, sorry. But I chose a sample from my data and using complete.cases function, eliminate observations with NA values. It worked nice.(I decided not to consider all of my data) thanks again. – user3519155 Feb 04 '16 at 12:05
  • I am trying to apply fuzzy clustering by different distance measurements on this likert scaled data. And I thought using "Percent Disagreement" method would be different because this method is useful if the data is categorical in nature. And this method is like D(x,y) = (number of xi not equal yi)/Nd . But I have 7 columns and 300 rows. How can apply this method by using R? – user3519155 Feb 04 '16 at 12:17

0 Answers0