MS Stats UChicago vs AM Stats Harvard (vs MS Stats Stanford)

[machinescholar]

Hi everyone,

 

I would appreciate if you may give me any advice on this.

I am a CS major undergraduate and am interested in doing theoretical research in Statistical Machine Leaning. While I did not get into the Stats PhD programs, I did got admitted into MS Stats at UChicago and AM Stats at Harvard. I also applied to Stanford but have not heard any thing from them so far.

I know Stanford and Harvard rank #1 and #3 in USNews in Statistics. However, I heard that some negative things about the Stanford programme that they admit too many Masters students (~100). While Harvard is selective, it is more geared for the industry. UChicago is lesser ranked (#5) but it looks like a solid PhD preparation programme.

Given that my future plan is to apply for PhD in Statistics a year later, should I choose UChicago over Harvard or Stanford (if I get admitted) ? Or will the brand recognition be more important in the long run ?

 

Thank you guys !

Edited February 7, 2017 by machinescholar

[Robbentheking]

I've done a good amount of research on masters programs. It seemed to me like Chicago was the best for PhD preparation in the country, hands down. I think Harvard and Stanford may be better for industry, both based on name recognition and the focus of the programs. Stanford I recall labels their program as a 'terminal masters' and specifically stresses that it is rare for masters students to obtain admission to the PhD program. Chicago seemed a lot more flexible/theory orientated. They've also taken a couple kids from the masters program as PhD candidates the past few years. Just a couple of data points.

I think Harvard was only a year long program? That probably isn't too helpful for you, as won't you application won't have substantially changed by next fall. Think about recommendations...you can't expect someone to write you a great rec after being in their class for 2 months.

To be honest though, I don't know anything about theoretical machine learning. Maybe you should be in a certain place just based on that interest. If it's going to be CS heavy, maybe you should be looking at school with a good CS department as well? Chicago doesn't really fit the bill there, as far as I know.

[machinescholar]

3 hours ago, Robbentheking said:

I've done a good amount of research on masters programs. It seemed to me like Chicago was the best for PhD preparation in the country, hands down. I think Harvard and Stanford may be better for industry, both based on name recognition and the focus of the programs. Stanford I recall labels their program as a 'terminal masters' and specifically stresses that it is rare for masters students to obtain admission to the PhD program. Chicago seemed a lot more flexible/theory orientated. They've also taken a couple kids from the masters program as PhD candidates the past few years. Just a couple of data points.

I think Harvard was only a year long program? That probably isn't too helpful for you, as won't you application won't have substantially changed by next fall. Think about recommendations...you can't expect someone to write you a great rec after being in their class for 2 months.

To be honest though, I don't know anything about theoretical machine learning. Maybe you should be in a certain place just based on that interest. If it's going to be CS heavy, maybe you should be looking at school with a good CS department as well? Chicago doesn't really fit the bill there, as far as I know.

Thank you for the advice ! 

Yes, Harvard 1 year duration and no thesis requirement are my biggest concerns. Also, at least from the past projects which I worked on, Theoretical Machine Learning is, in my opinion, more closely related to Maths and (Theoretical) Stats than CS.

 

Edited February 7, 2017 by machinescholar

[Severina]

All three are top programs from which you can get a good job or get into a top PhD program (assuming you get good grades).  Focusing on whether the school in question is ranked #1, #3, or #5 is slicing the baloney a little too thin.     Pretty much any top ten program can take you wherever you want to go.  Focus instead of the content and requirements of the individual programs.  

[Robbentheking]

39 minutes ago, machinescholar said:

 

Yes, Harvard 1 year duration and no thesis requirement are my biggest concerns. Also, at least from the past projects which I worked on, Theoretical Machine Learning is, in my opinion, more closely related to Maths and (Theoretical) Stats than CS.

 

 Obviously you're not doing a PhD (yet), but maybe glancing at the actual work that faculty are doing wouldn't be a bad idea. All the schools you mentioned have insane math departments, so you're not going to get much insight by looking at it on that level.

 

[Tigertiger1993]

Actually, the #3 in USNEWS is Harvard Biostat, yet Harvard Stat is rank #7, which is behind Chicago Stat. Maybe this information provides you more preference for Chicago's program.

[abstract_art]

7 minutes ago, Tigertiger1993 said:

Actually, the #3 in USNEWS is Harvard Biostat, yet Harvard Stat is rank #7, which is behind Chicago Stat. Maybe this information provides you more preference for Chicago's program.

Just to add to this here are the US News rankings split up between stat and biostats:

 

[GoPackGo89]

5 hours ago, marmle said:

 

 

Any guess when these rankings will be updated? Ever since I've referenced that list it has been from 2014 so I am not sure how often it is updated

Edited February 10, 2017 by statbiostat2017