Berkeley statistics phd vs NYU data science phd

[Prestvana]

I was set on going to berkeley but I was just recently notified about NYU. Its a whole confusing thing but i still want to consider NYU.  Anyone have any thoughts between these two programs in terms of job placement in industry or just general stuff? My research interests are pretty flexible, gravitating more towards CS/stat topics like ML. I visited berkeley but not NYU. If anyone who visited NYU's program could chime in it would be helpful. 

[Wzz]

I visited NYU CDS and quite liked it. I feel like CDS differs very much from those traditional stats programs, let along Berkeley which is very theoretical among stats. So here is some ideas I got from visiting:

1. This is a new program, so people are very passionate. The center has great power to allocate resources, and basically will be very personalized to student's path. This all make sense, as they really want the first year to be a success. Support will be strong, and I suppose they will do whatever they can to make sure this group of student have good placements. Professors are all hired in CDS and another related department (linguistics, social science, math, etc), and promise to spend half of time in CDS. This is a good portal to interact with other departments too. And CDS seems to have a lot of funding. 

2. Very good links with industry. They mentions many times that they expect students, if not continuing research during summer, to do summer intern in the companies in NYC. It seems to me that those companies (+facebook maybe) are looking forward to this batch of students to do data science stuff in industry.

3. Flexibility.  For example, they said the course system is open to changes. Basically everything is new so many details are planned to be figured out on the run. And the atmosphere is very chill. 

(4. Though it is in NYC, the stipend w/o TA burden + summer internship would provide better living quality comparing with some other schools. No TA burden is very appealing to me, as it would save much time.)

A few thought of mine:

1. The core people at CDS are doing different things (optimization/NLP/ML in social sciences/urban/visualization), and I feel like I would come here only if I am super interested in one of these fields (but not traditional statistics). I am not sure what people would get out of the main courses if coming without a clear idea which facet of data science to jump into. 

2. Research. Their research are very interesting, and covers many topics that fascinates me. But it would be totally different from math/stats research. They reminded me that stats research is not the same as data science. 

3. This is a new program. Alongside with the benefit of being in a new program is the risk of getting lost, especially when this program is trying to define data science as a subject itself. They say even they don't have a clear definition for data science, and hope to figure it out as the program goes. It is exciting to explore this not-well-defined field with this group, but some (like me) may want to follow a more clear path. 

In general, I love this program and recommended many people to apply next year, as I can totally see them becoming the leading program in data-science phd. But comparing with statistics, it would be another story. I don't know what the need in academia would be for data science people in five years. And the factors on which to choose between CDS and any stats program is simple: what kind of research you want to do, and what kind of job you wish do after the phd. 

Turning down this offer is painful, as I can only imagine how happy and excited I will be if I come, but I chose to be conservative in making this choice. 

[Prestvana]

Thanks a lot for the response! The NYU program seems fascinating and I wish I had the chance to visit. It will be really hard for me to turn this offer down as well. 

I am leaning towards Berkeley for the following reasons. 1. I want to be in CA during PhD and after. 2. Although I'd be situated in the stat department, the program seems very flexible in terms of course requirements and advisers. Berkeley stood out in this regard, compared to the other schools I visited. I was told I could be advised by professors in stats or related departments. I am sure NYU is similar as well in this regard due to DS being so interdisciplinary 3. Berkeley stat is already an established program with many well-known professors and alumni seem to be doing really well in terms of jobs in industry/academia. I was told it wasn't hard to obtain internships since a lot of companies try to recruit Berkeley students 4. I had a chance to visit Berkeley during their visitation days and liked the school/people/program a lot so I have a better idea what to expect. 

However, NYU's offer in terms of its stipend is much more attractive than Berkeley. I wish NYU had let me know a month ago whether I had been accepted and not a couple of days before April 15. I was under the impression that I was not going to be considered, especially since I was not invited for their visitation event back in February.

[Wzz]

11 hours ago, Prestvana said:

Thanks a lot for the response! The NYU program seems fascinating and I wish I had the chance to visit. It will be really hard for me to turn this offer down as well. 

I am leaning towards Berkeley for the following reasons. 1. I want to be in CA during PhD and after. 2. Although I'd be situated in the stat department, the program seems very flexible in terms of course requirements and advisers. Berkeley stood out in this regard, compared to the other schools I visited. I was told I could be advised by professors in stats or related departments. I am sure NYU is similar as well in this regard due to DS being so interdisciplinary 3. Berkeley stat is already an established program with many well-known professors and alumni seem to be doing really well in terms of jobs in industry/academia. I was told it wasn't hard to obtain internships since a lot of companies try to recruit Berkeley students 4. I had a chance to visit Berkeley during their visitation days and liked the school/people/program a lot so I have a better idea what to expect. 

However, NYU's offer in terms of its stipend is much more attractive than Berkeley. I wish NYU had let me know a month ago whether I had been accepted and not a couple of days before April 15. I was under the impression that I was not going to be considered, especially since I was not invited for their visitation event back in February.

They gave me the impression that they really want to make sure the quality of the first batch of students, and really don't want to over-enroll. Among the interviewed students (~20) there were even many who didn't get offer. I turned them down early to make room for other people, but I know many others may take a long time to decide. I think they might have over-estimated the acceptance rate when only inviting ~20 people to visit. So, yeah I agree they might lost many potentially interested student in the long process of waiting.