I have a data set

`> head(data) id centre u time event x c1 c2 c3 c4 c5 c6 c7 c8 c9 c10 c11 c12 c13 c14 1 1 0.729891 0.3300478 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 2 1 0.729891 7.0100000 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 3 1 0.729891 7.0150000 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 4 1 0.729891 1.3616940 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 5 1 0.729891 7.0250000 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 6 1 0.729891 5.0824055 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 `

and I want to fit a proportional hazards model.

**IN R**

I first defined `formula`

:

`> formula Surv(time, event) ~ x + c1 + c2 + c3 + c4 + c5 + c6 + c7 + c8 + c9 + c10 + c11 + c12 + c13 + c14 `

and I used the `coxph`

function:

`> library(survival) > mod <- coxph(formula, data=data) Erreur dans fitter(X, Y, strats, offset, init, control, weights = weights, : NA/NaN/Inf dans un appel à une fonction externe (argument 6) De plus : Message d'avis : In fitter(X, Y, strats, offset, init, control, weights = weights, : Ran out of iterations and did not converge `

**IN SAS**

I used `proc phreg`

:

`proc phreg data=data; model time*event(0) = x c1-c14 / ties=breslow; run; `

Partial output:

` Convergence Status Convergence criterion (GCONV=1E-8) satisfied. `

**Contents**hide

### Question:

- Do you have any suggestion to make it work in R?

I have tried to increase `iter.max`

but the problem is still there… If needed, I can provide the data (600 rows).

#### Best Answer

What about changing of the starting values? Supply to `init`

similar values to the output of SAS and see what happens. This in general is not a good strategy to get the convergence, but it helps to see whether the problem is in algorithm, or just in starting values.

### Similar Posts:

- Solved – How to get convergence using coxph (R) given that model converges using proc phreg (SAS)
- Solved – How to get convergence using coxph (R) given that model converges using proc phreg (SAS)
- Solved – R: CoxPH model with a categorical variable that has too many levels
- Solved – R’s coxph won’t converge when I include factor (categorical) variables
- Solved – What are the options in proportional hazard regression model when Schoenfeld residuals are not good