# regression approach for missing data (left censoring?)

I have a regression problem where I want to predict actuals (dependent variable) of some process where I only have values for a small number of independent variables at the beginning of the process but more non missing data as the process matures. This is kind of similar to survival analysis, which I have used in the past, but with left censoring. However, in contrast to survival analysis I want to predict the value of a continuous variable instead of a (survival) probability. Are there specific regression approaches I can use (e.g. truncated regression, censored regression)? Any pointers, especially to R/Python packages, would be very much appreciated. Many thanks!

• I’m not sure I completely understand your setting, but for censored regression you can use the survreg() function from the survival package in R. – Dimitris Rizopoulos Oct 13 '18 at 21:05
• Could you please tell me what is not clear so that I can improve my question? Btw can survival analysis not only be used to predict hazards/probabilities. In my case, I would like to predict a continuous variable as in regression ... – cs0815 Oct 14 '18 at 10:19
• I also cast the problem differently here: stats.stackexchange.com/questions/371675/… – cs0815 Oct 14 '18 at 10:25
• The survreg() function can be used for censored regression. Hazard ratios and survival probabilities are more natural in the Cox models framework fitted by coxph(). Though, of course, because the hazard function is linked to the density and survival functions. Even if you fit the model with survreg() you could, in principle, calculate the hazard. – Dimitris Rizopoulos Oct 14 '18 at 14:53