search

bhklab / RadioGx

RadioGx and PharmacoGx: Integrated prediction tools for combining radiotherapy with systemic agents
5 stars 7 forks source link

Error with the residual distributions, Cauchy vs. Normal #32

Closed vmsatya closed 7 years ago

vmsatya commented 7 years ago

We obtain completely different alpha values. Below is the synthetic data: linearQuadraticModel(c(0,1,2,3),c(1.1, 0.2, 0.7, 0.1),SF_as_log=FALSE,family="normal") $alpha [1] 0.3 $beta [1] 0.1322388

linearQuadraticModel(c(0,1,2,3),c(1.1, 0.2, 0.7, 0.1),SF_as_log=FALSE,family="Cauchy") $alpha [1] 0 $beta [1] 0.255527

vmsatya commented 7 years ago

I get different values now for the parameters:

linearQuadraticModel(c(0,1,2,3),c(1.1, 0.2, 0.7, 0.1),family="normal") $alpha [1] 0.6489688

$beta [1] 0.00403125

linearQuadraticModel(c(0,1,2,3),c(1.1, 0.2, 0.7, 0.1),family="Cauchy") $alpha [1] 0

$beta [1] 0.2555312