Mixture of Censored iid Normals

Ok thanks! The reason why I needed imputation is more experimental carelessness in this case so I would more prefer it to be resolved experimentally but wanted to see in any case if I could do something on the computational side.

ps: I am also aware of things like iterated k-means to “guess” missing values whose generalization to here would be trivial but the way missing values are distributed in my case would make it not a reasonable approach.