Master's Statistics: Columbia v/s Duke

[StarPower59]

Hi,

If getting a job in companies like IBM,microsoft, amazon (basically tech) is the end goal of studying a program, which program is better?

 

Thanks!

[machinescholar]

7 hours ago, StarPower59 said:

Hi,

If getting a job in companies like IBM,microsoft, amazon (basically tech) is the end goal of studying a program, which program is better?

 

Thanks!

It would help if you specify which type of jobs from these companies you want to land. In general, I think either of them would not help that much. if you are aiming for a Data Scientist or Quant Analyst role, it is very challenging because these top techs mainly recruit from the PhD graduate pool. As such, Masters candidates are mostly placed into Software Engineering Positions where the coding interviews are all that matters...

Though between Duke and Columbia, I would pick Duke Stat MA programme as the cohort is much smaller and the placement record is quite strong in respect to a Masters program:

http://stat.duke.edu/people/masters-alumni-pages

Edited March 6, 2017 by machinescholar

[Innominate]

Duke's MS has about 25 students per year (currently 51 for a two year program) and Columbia's has 160+ per year.  160 students per year just screams cash cow to me, but you can decide for yourself.  I'm not aware of any other program that even approaches the size of Columbia.

Duke has an extremely strong focus on Bayesian statistics, which lends itself to some machine learning work as well.  I know less about Columbia but they do tons of social science work and I think they tend to place people into financial positions generally.  

Also duke has unusually clear information about MS grads so scroll through them and see what you think of where they're placing (same link as what machine scholar posted)

http://stat.duke.edu/people/masters-alumni-pages

[poopyhead]

3 minutes ago, Innominate said:

Duke's MS has about 25 students per year (currently 51 for a two year program) and Columbia's has 160+ per year.  160 students per year just screams cash cow to me, but you can decide for yourself.  I'm not aware of any other program that even approaches the size of Columbia.

Duke has an extremely strong focus on Bayesian statistics, which lends itself to some machine learning work as well.  I know less about Columbia but they do tons of social science work and I think they tend to place people into financial positions generally.  

Also duke has unusually clear information about MS grads so scroll through them and see what you think of where they're placing (same link as what machine scholar posted)

http://stat.duke.edu/people/masters-alumni-pages

Given that Andrew Gelman is at Columbia, I'd expect that they're very much Bayesian. 

That said, if you're one of 320 MS students (assuming it's a two year program), good luck getting any time with him.

Edited March 7, 2017 by poopyhead

[machinescholar]

2 hours ago, poopyhead said:

Given that Andrew Gelman is at Columbia, I'd expect that they're very much Bayesian. 

That said, if you're one of 320 MS students (assuming it's a two year program), good luck getting any time with him.

As a Masters student, you wont have a chance at all. Big names like Gelman and Blei only teach PhD classes and unless you are very top MS students, it is difficult to even seek for permission to attend them.

On the other hand, the MS curriculum is actually very balanced between Frequentist and Bayesian Statistics.

Edited March 7, 2017 by machinescholar

[StarPower59]

13 hours ago, machinescholar said:

It would help if you specify which type of jobs from these companies you want to land. In general, I think either of them would not help that much. if you are aiming for a Data Scientist or Quant Analyst role, it is very challenging because these top techs mainly recruit from the PhD graduate pool. As such, Masters candidates are mostly placed into Software Engineering Positions where the coding interviews are all that matters...

Though between Duke and Columbia, I would pick Duke Stat MA programme as the cohort is much smaller and the placement record is quite strong in respect to a Masters program:

http://stat.duke.edu/people/masters-alumni-pages

I would like to work as a data scientist and I want to get into data industry with no interest in PhD. I am confused between these two because -

1. Location: Columbia is in new york and thus a better place to be in terms of networking and opportunities while Durham is a small place.
2. Course structure: Duke's structure seems more serious and focused and would serious data companies fancy that as compared to Columbia?

3. Thesis & internship: Duke has a thesis, as well as internship requirement while Columbia doesn't. Does thesis matter to recruiters?

Thanks a lot!

[edward130603]

Duke is located close to NC's Research Triangle Park region, which has companies like IBM and Cisco. As for the other two, which employer wouldn't want a more serious degree?