search

HBGDki / hbgd

hbgd
http://HBGDki.github.io/hbgd/
MIT License
1 stars 1 forks source link

can't find example with sitar that works #5

Open schloerke opened 8 years ago

schloerke commented 8 years ago

reaches singularity point

get_fit(cpp, y_var = "haz", method = "sitar")
Error in solve.default(pdMatrix(a, factor = TRUE)) : 
  system is computationally singular: reciprocal condition number = 3.98921e-21
In addition: Warning messages:
1: In nlme.formula(y ~ fitnlme(x, s1, s2, s3, a, b, c), fixed = s1 +  :
  step halving factor reduced below minimum in PNLS step
2: In nlme.formula(y ~ fitnlme(x, s1, s2, s3, a, b, c), fixed = s1 +  :
  step halving factor reduced below minimum in PNLS step
hafen commented 8 years ago

Try with y_var = "htcm". We need to add a check to sitar that makes sure you aren't fitting on the z-score scale, since it's not designed for that. Even on the original scale, it is difficult to converge.