My purpose of paper is to show that sometimes correlation between parameters can be by chance and not by causality.

 

Definition:

Pearson correlation coefficient (usually called merely ‘correlation coefficient’) measures a linear relationship between two variables. It’s worth noticing Pearson correlation coefficient reflects an extent of linear relationship (not exponential or quadratic relationship).

The range of the correlation coefficient is [-1.0,…,1.0], the more the coefficient close to 1 or -1 the more linear relationship found between two variables. If the coefficient is zero or nearby then no linear correlation is detected.

 

Use case from video compression

You are required to estimate a linear relationship between a scene complexity (e.g. the percentage of edge pixels) and the generated frame size, where an encoder is configured with constant QP (otherwise if Rate Control is applied frame sizes are similar in magnitudes. In case with Rate Control it’s worth checking a relationship between video quality and complexity).

You expect the following: more complex scene more bits are produced.  To check the relationship  you choose 10 clips (or scenes), the complexity of each scene is pre-computed. Then you apply the encoder on each excerpt and calculate Pearson correlation coefficient and plot the scatter graph (schematically):

Let’s assume that the correlation coefficient r = 0.5.

Now the question is raised – maybe, no linear relation exists and we get r=0.5 by chance?

What’s a confidence (or a probability) of getting r=0.5 provided that no linear relationship exists?

Or in statistics jargon, is the correlation coefficient r=0.5 is significant or not? 

 

Significance testing of Pearson correlation coefficient is based on t-student distribution with (n-2) degrees of freedom, where ‘n’ is the number of samples (in our case 10):

We test two hypothesis:

 

  • Null Hypothesis: complexities and frame sizes are linearly unrelated, i.e. r = 0

 

  • Alternative Hypothesis: complexities and frame sizes are in someway related, r ≠ 0 (two-sided) or r>0 (one-sided, i.e. a positive linear relationship conjectured).

 

So, for our case with r=0.5 and n = 10 (8 degrees of freedom) we get t-value ~1.64.

According to t-student table for 8 degrees of freedom the confidence level that there is a linear relationship between complexities and frame sizes is around 85%:

Usually the confidence of 85% is not sufficient to reject the null hypothesis. It’s common rule: the null hypothesis can be rejected if the confidence is above 95%

Even if we consider the alternative hypothesis with positive linear relationship the confidence is ~93%.

 

Conclusion

The confidence 85% of a linear relationship between complexities and frame sizes is not sufficient to assert that a correlation between these two variables exist.

18 Responses

  1. With havin so much content do you ever run into any issues of plagorism or copyright infringement? My site has a lot of completely unique content I’ve either authored myself or outsourced but it appears a lot of it is popping it up all over the internet without my authorization. Do you know any techniques to help protect against content from being stolen? I’d genuinely appreciate it.

  2. I have been browsing online more than 3 hours today, yet I never found any interesting article like yours. It’s pretty worth enough for me. In my view, if all website owners and bloggers made good content as you did, the net will be much more useful than ever before.

  3. I was suggested this website by means of my cousin. I am no longer sure whether or not this put up is written by way of him as nobody else know such exact about my difficulty. You are wonderful! Thanks!

  4. I’ve been exploring for a bit for any high quality articles or blog posts on this kind of area . Exploring in Yahoo I at last stumbled upon this website. Reading this info So i’m happy to convey that I’ve a very good uncanny feeling I discovered exactly what I needed. I most certainly will make certain to don’t forget this web site and give it a look regularly.

  5. My developer is trying to persuade me to move to .net from PHP. I have always disliked the idea because of the expenses. But he’s tryiong none the less. I’ve been using Movable-type on a variety of websites for about a year and am nervous about switching to another platform. I have heard excellent things about blogengine.net. Is there a way I can transfer all my wordpress content into it? Any help would be greatly appreciated!

  6. I like what you guys are up too. Such intelligent work and reporting! Carry on the superb works guys I have incorporated you guys to my blogroll. I think it’ll improve the value of my site 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *