Data Science Blues

Started masters program for comp sci, hoping to be a data scientist.

The class I’m currently taking has me working in scala, pytorch, mlib, hadoop, hive, pig, and pandas.

I feel like I’m barely surviving. I thought I liked the idea of being a data scientist because I enjoy coding and doing analytics at my job, but this is really something else and if this is what I can expect for the rest of the program + my future job, I’m really having doubts that this is for me.

I guess I should just be an analytst?

Is there a version of data scientist that exists that doesn’t require intense amounts of calculus and linear algebra? How about just importing ML libraries in python and calling the pre-built functions, tuning parameters, and then examining the results without all of the math nonsense?

Is the term for that stupid data scientist? Data engineer? Someone with experience please chime in.

You just apply to the same data science jobs and say you’re self-taught

1 Like

Real life jobs use a lot less calculus (and really hard math in general) than what you learn in school.

You might talk to people who actually do the job to see how math-intensive it really is.

1 Like

the people who go to school for comp sci are the people who end up not working in the field after graduation

why would someone working with computers and logic hire someone who chooses the inoptimal choice of paying thousands of dollars to get a job when they can learn it for free and get the same job

I truly feel sad you decided to go to school for comp sci and now you’ve limited your job prospects to real estate and general contracting

that sounds like really low skill work.

keep in mind that 4-5 years ago, to accomplish the same tasks would require in-depth knowledge about “ML” and you’d most certainly have to fnagle with an underlying codebase. now it’s 30~ lines of python a CS sophomore could do.

you should brand yourself as a “data engineer” and learn skills to do the following: sanitize, order, and store datasets

it’s all just going to be compressing a csv file and uploading it to GoogleML Cloud in 3 years anyway.

reader’s note: this poster is not employed.

2 Likes

Google search for libraries to do the following:

When you are computing in the cloud, like me (i store all my macros in Dynamo DB) you do not have to think about writing efficient, clean code. You can just add resources until everything works for the low, low price of 4¢ + a quick blowie in the wash closet for jeff b

this is actually completely incorrect

1 Like

all the friends i knew who got employed at facebook/google/amazon/salesforce were compsci majors

sample size like a dozen or so

and most compsci majors i knew are still in the field

idk about grad school for data analyst or anything tho

Keep in mind I don’t actually have any idea what im talking about i just post for the sake of hearing myself talk

1 Like

yeah there’s going to be no library to parse 5 million json snippets coming off IOT refrigerators from your third-party MQTT broker

jones talks a big game spewing off buzzwords like a newly trained salesman but the depth of his knowledge is just a small puddle on the street which is why he is unemployed to this day

1 Like

jones being the 6 foot 7 freak he is would be better suited as an nba player than a computer guy

I could probably do that tonight in 15 minutes using AWS Web Services

do you think there is a market for smart refrigerators scanning barcodes and selling customer eating habits to advertisers

5 million json snippets = 5 million aws lambda endpoints each individually running the parsing code which i copied and chopped together from 4 different stack overflow answers