First I know that the Compound Poisson process, is the sum of i.i.d random variables X1,X2,… and we have a counting process {Nt}, I have checked your code and Nt is not a counting process because the property of this process is Nt<Ns if t<s.

Also I am not clear with Nt, It should be a poisson process, but i found some code in R on internet but most of them are not a counting process. I don´t understand why all the numbers given are non integers.

Regards.

]]>I need help understanding the Residual vs Actuals in relation to the Residual vs Fit plot. (I feel that one can be obtained from the other, but not clear how).

I am using the equation e = y -yhat,

where e=residual,y=actual, yhat=fit (i.e. predicted)

I have Tobit model with ‘y’ censored to lie between [0,1].

The Residual vs Actual plot is roughly an upward trending line- Residuals are on the Y-axis and Actuals on the X-axis.

Here is a rough table of the data:

For a fixed value of y, say:

(1.) y=0,the band of residuals is between -0.25 and -0.1

(2.) y=0.2, the band of residuals is between -0.4 and 0

(3.) y=0.4, the band of residuals varies from -0.3 to 0.1

(4.) y=0.6, the band of residuals varies from -0.2 to 0.4 and

( 4.) y=1, the band of residuals varies from 0.25 to 0.75.

How can I tell that the Residuals vs Fit will be a downward sloping line (confirmed by SAS)? I think I should be able to just by looking. Note: The Fit values lie in about (0.05,0.75).

Thanks for any help!

]]>Do you know any techical articles that study this relationship between the Expected Shortfall and the Parametric VAR?

I read the Swiss report you mentioned, but they don’t give any explanation on these numbers.

Thank you,

Leticia

]]>I suspect that you are correct for a genre like electronic music, which basically started as a genre in the mid-1970s, and which, to the best of my knowledge, reached its highest BPM peak in the mid-1990s, under the influence of several strands of techno (acid and trance in particular). Note that electronic music is an interesting case study here, because the BPM is part of the information used to compose and select the music.

Disconest might also help:

http://www.disconest.com/?discogsurl=https://www.discogs.com/release/8698126

]]>You cannot listen to 2015 CDs in 1990 while you may be able to listen to CDs purchased in 1990 or remastered from 1990) in 2015.

Obviously with count data of actual listens this becomes less relevant except that CD with Despacito on it, you don’t listen to because it is always on your car radio 🙂

]]>One suggestion though: you would probably avoid any supecious bumps (if any) by using the np package and a cross-validated bandwidth . Moreover, you would be able to get (and to plot) the first derivative without any additionnal effort. That would certainly confirm your feeling but would be a more robust result.

Anyway nice post! ]]>

and in addition wealth and income are different and though highly correlated can be distorting! ]]>

On pourrait conduire cette métaphore de manière intelligente, en réalisant une valorisation en journées de travail des contributions payées (sur le cycle de vie), et des prestation et services reçus (idem). L’écart entre les pays mesurerait alors des différences d’efficacité des systèmes en jeu.

]]>If it is the case we have a potential issue as I can’t install the winference stuff with devtools package. Did I miss anything? ]]>