Question

I’m working with the fractal package and using the DFA function which gives me a very complete output in the console :

 DFA(log(price_1m[1:1024]),detrend="poly2",overlap=0,scale.max=512)

Console output :

Detrended fluctuation analysis for log(price_1m[1:1024])
--------------------------------------------------------
H estimate             : 0.452347 
Domain                 : Time 
Statistic              : RMSE 
Length of series       : 1024 
Block detrending model : x ~ 1 + t + I(t^2) 
Block overlap fraction : 0 
Scale ratio            : 2 

Scale 6.00000000 1.2000e+01 2.4000e+01 4.8000e+01 96.0000000 1.920e+02 3.8400e+02
RMSE  0.00029362 4.2002e-04 5.2302e-04 8.5258e-04  0.0011629 1.403e-03 1.7068e-03
Message d'avis :
In if (cls == "signalSeries") { :
  la condition a une longueur > 1 et seul le premier élément est utilisé

the fact is thah I only want and need the Hurst exponent estimate. So I would like to get rid of all the rest.
So I’ve tried to get the attributes of the DFA function (the DFA function give a fractalBlock object) and here is what I get in the console :

Call:
lm(formula = y ~ 1 + x, data = data.frame(x = xx, y = yy, w = ww), 
    weights = w)

Coefficients:
(Intercept)            x  
    -8.9269       0.4523  

Message d'avis :
In if (cls == "signalSeries") { :
  la condition a une longueur > 1 et seul le premier élément est utilisé

Ok, I’ve got a lm object from which I can extract the slope parameter estimate with :

attr(DFA(log(price_1m[1:1024]),detrend="poly2",overlap=0,scale.max=512),which="logfit")$coef[2]

and here is what it’s printed in the console :

       x 
0.452347 
Message d'avis :
In if (cls == "signalSeries") { :
  la condition a une longueur > 1 et seul le premier élément est utilisé

and here is the point : I only want to get the numeric value that is 0.452347. I’ve tried to convert my data using the as.numeric() function but it does not work.

Thanks for your help.

Answer 1

This would give you an unnamed Double as a result:

attr(DFA(log(price_1m[1:1024]),detrend="poly2",overlap=0,scale.max=512),which="logfit")$coef[[2]]

But would maybe still include the warning. What if you first assign to a variable and then print?

x <- attr(DFA(log(price_1m[1:1024]),detrend="poly2",overlap=0,scale.max=512),which="logfit")$coef[[2]]
x

If you still receive this warning you can perhaps suppress it with try:

try( x <- attr(DFA(log(price_1m[1:1024]),detrend="poly2",overlap=0,scale.max=512),which="logfit")$coef[[2]], silent = TRUE)
x