How to interpret numerical results from Eucledean distance

How to interpret numerical results from Eucledean distance

16.07.2018, 10:54

I currently run analyses for document similarity and I need to use a code
frequency analysis in Similarity for Documents analysis, either block or eucledean. The problem is how to
interpret these numbers. How do I know a value of 120 is good or bad? Is
it possible to normalize the values between 0 and 1?



Thanks and regards,

Mauricio

Version: MAXQDA 2018
System: Windows 10
mauricebar
 
Posts: 3
Joined: 12.07.2018, 15:13

Re: How to interpret numerical results from Eucledean distan

17.07.2018, 15:32

Dear Mauricio,
thank you for your question and welcome to the MAXQDA Forum!

I've asked our expert on this feature, but he is currently on vacation. Have you read the respective entry in our manual? You can find it here:

https://www.maxqda.com/help-max18/mixed-methods-functions/similarity-analysis-for-documents

Maybe this answers your question? If not, I'll be happy to pass on the answer of our expert on this issue.

Kind regards,

Andreas
MAXQDA Support Team
Andreas V.
 
Posts: 272
Joined: 13.04.2017, 16:23

Re: How to interpret numerical results from Eucledean distan

17.07.2018, 21:48

Dear Andreas,

Thanks for your response. Indeed, I have checked the manual but I cannot find that information there. I am aware it is something a bit more Mathematics related, but it also has to do with how the software presents the data, I believe. I would certainly be really glad to learn what your colleague can contribute.

Thanks and regards,
Mauricio.
mauricebar
 
Posts: 3
Joined: 12.07.2018, 15:13

Re: How to interpret numerical results from Eucledean distan

21.07.2018, 16:26

Hi Andreas,

I was wondering if you had any luck with the expert about this topic. It would help a lot to have some guidance here.

Thanks and regards,
Mauricio
mauricebar
 
Posts: 3
Joined: 12.07.2018, 15:13

Re: How to interpret numerical results from Eucledean distan

24.07.2018, 13:10

Hi Mauricio,

yes, he's back and I have an answer:

You can convert a similarity matrix into a distance matrix with values between 0 and 1. For each cell, you need to calculate the following:

distance = 1 – similarity / max(all similarities)

This will standardize the values in regard to the greatest similarity in the matrix. So this is a "relative", not an "absolute" assesment of similarities.

I hope this helps! In case of any further questions, don't hesitate to ask.

Kind regards,

Andreas
MAXQDA Support Team
Andreas V.
 
Posts: 272
Joined: 13.04.2017, 16:23

Return to Technical Questions

Who is online

Users browsing this forum: Google [Bot] and 8 guests

We use cookies to improve your experience on our website. By clicking OK or by continuing to browse the website, we’ll assume that you are happy with their use. Click here to review our Cookie Policy. OK