nlpr

Thanks for you helpful comment Elephas. I agree with you in this respect, but was concerned that statistics programs may expect to see someone speaking to my mathematical background/maturity (my other two LORs are from CS profs), sort of like how they expect analysis on my transcript. I'll have to hang around the forum more and check out some strong statistics applications. Looks like you're starting your PhD at NU--congrats and good luck!

I'm a CS undergrad from an average state school applying to statistics programs this fall. Earlier my interests were more focused on AI/NLP/speech, so I volunteered at a EE professor's speech processing lab for several semesters; it was mostly mundane work like data collection (including designing a web interface for users to listen to/evaluate speech samples etc.), nothing really considered "research" in the field. Recently I've been self-studying some real analysis/stochastic processes/Bayesian statistics and reading more theoretical work, and decided statistics PhD is probably a better fit for me. Does it still make sense to have a LOR from my EE professor? BTW he is senior and established in the speech processing field; due to how busy he is I've interacted more with his grad students. I asked him in person and he said he's happy to write me a letter. I'm also taking Real Analysis 1 this semester, and I've been wondering if I should visit my analysis professor's office hours more and get a LOR from him instead. He's a young tenure-track assistant prof doing work in dynamical systems. (I've self studied through measure theory so Analysis 1 will be an easy A for me; I'm taking it b/c it seems many statistics PhD programs require analysis on transcripts) The problem here is of course he may not know me for very long by November/December.