0
$\begingroup$

I'm trying to deskew some distributions based on checks that verify Skew and Kurtosis and within range, if not, I was going to attempt to deskew the data.

On wikipedia, there's a mathematical notation for the CDF of a skewed distribution.

Could someone show me how this would be setup in excel?

http://en.wikipedia.org/wiki/Skew_normal_distribution

Also, anyone know how this varies from a Box Cot estimated lambda distribution transformation?

$\endgroup$
1
$\begingroup$

I don't think this version of skew normal will work for you. You already have some data, and you'd like to transform them to normality? Then Box-Cox is probably the first thing to try, but the data should be positive-valued. Also, you cannot fix both skew and kurtosis with these, only skew.

In fact, I wouldn't rely on skew alone, but make histograms of your transformed variables. Excel histograms are notoriously bad, maybe it's time you tried R.

| cite | improve this answer | |
$\endgroup$
1
$\begingroup$

If we want to plot the skew normal curve in Excel, we would use something like this:

The skew distribution curve is $$f(\alpha,x)=2 \phi(x) \Phi(\alpha x)$$

Where (in almost-Excel) $$\phi(x)=normal~distribution(x)=norm.dist(x,0,1,False)$$ and $$\Phi(x)=cumulative~normal~distribution(x)=norm.dist(x,0,1,True)$$

So combined, we get (still in almost-Excel) $$f(\alpha,x)=2*norm.dist(x,0,1,False)*norm.dist(\alpha*x,0,1,True)$$

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.