• 瀏覽: 3,195
  • 回覆: 9
[隱藏]
如題
而且讀完呢科之後可以出嚟做咩工
發展方向同埋出路如何



回覆 引用 TOP

熱賣及精選
I will bite. To give you some perspective where I am coming from ... i teach both master/PhD students data science techniques (my class focuses on stochastic modeling in R), do research in AI (with 2 PhD students and a colleague) and I advise and consult for silicon valley companies (2 start-up + 1 big company) on both data science and behavioral economics issues. 


Data science is a broad field dealing with data issues. Typically, companies hire data scientists from a multitude of disciplines including CS, information systems, economics (mostly econometrics), and OR, in addition to "pure" data science programs.


Methodologies used are also very broad from the CS-centric data mining/machine learning techniques like random forecast, to the stat/econometrics type techniques like the many variations of regression & time series models, to the modern deep learning neural network. Obviously I cannot teach the details of any of these here. But I will point out one key issue to give you a sense. The CS ML/AI techniques are good at recognizing patterns, and predict from those patterns. However, they are bad at identifying cause and effects and figure out WHY something happen (or how the system will respond to changes), where some of the econometrics techniques are better at. The extreme case is that even if a AI/DLN performs very hard, it is challenging to figure out why and how it does it (alpha go is a good example). Personally i think combinations of these different methods are the future. I have one colleague whose early claim to fame is a combination of ML techniques to the very traditional propensity score ... to figure out causal response.


Data science is now broadly applied to many many applications. In business, the most obvious is marketing (digital marketing analyzing churns, purchases, weblogs, engagement and so on and so forth). The group i am consulting for is doing these kinds of work. However, these techniques are also applied to banking like evaluation of loans, fraud detection, stock analyses and a thousand other things. My own work takes AI into supply chain management analyzing human-decision making.


Jobs are plentiful. Both traditional large companies and blue-sky start-ups are hiring. I do not have stats right now but I can give you examples. My best (master, not PhD) student got hired by a cancer research group in the University of Chicago managing gene-related data. Another one was hired by a NYC start-up doing cyber-security analysis. This girl had, as i recalled, 3 offers .. the other two are more traditional places like a big insurance company. Another one went to a company making and maintaining seats on airplanes and her job is to analyze maintenance data (like inventory of parts) and make the business more efficient.


Let me point out one more thing. While the key of data science is to understand data and analyze business issues accordingly, the ability of dealing with data is important. Hence, knowing SQL and be able to program for data manipulation/cleaning will be useful for finding a job. For larger companies, they sometimes will get a data engineer to help you, but they always value people who are resourceful and can do it on their own.


Note that I am in the US so I know relatively little about HK. My sense is that while data science is needed everywhere (and there are a lot of companies here trying to catch up), HK may not be the forefront, and hence not the best place to pursue a data science career.  



引用:
原帖由 narius 於 2019-6-14 10:53 AM 發表

I will bite. To give you some perspective where I am coming from ... i teach both master/PhD students data science techniques (my class focuses on stochastic modeling in R), do research in AI (with 2  ...
Thank you for your sincere reply. I’m in the US either. I currently major in cog sci but I’m not really into it. I haven’t decided my dream job or career path but I wish I can work for FAGA in Silicon Valley. It is extremrly hard to declare my major as computer science, and one of my friend told me that the area of data science is closer to CS. However, I know nothing about it. Do you recommend I change my major to DS?



引用:
原帖由 Iamsillyman 於 2019-6-14 11:11 AM 發表


Do you recommend I change my major to DS?
I figure you mean CS? Without knowing your background and aptitude better, there is really no way for me to give you appropriate advice. However, I will point out 3 relevant issues that you can go sort out.

1) CS focuses more on the methods. If you like math and inner working of technical stuff, it is appropriate. However, the usual weakness of a CS trained data scientist is that s/he will have less knowledge of the application domain.

2) The alternative, assuming you are not veered too much away to economics or OR, is information systems (usually in a business school, although U of Michigan has a school of information). IS (or business analytics program run by an IS department like ours) is more focused on the application. You still get all the techniques, but you will also learn to use it in specific business problems. For example, my class projects are taken out of actual data science projects i did in the silicon valley. This distinction, BTW, is not 100% binary, obviously. If you really care a particular application (like marketing), you can go there too.

3) To be honest, a BS is not quite enough to make it in the data science world. When the corporate marketing data science team, i worked with, recruited last time, they won't even talk to master students and they were only interested in PhDs. And if you are planning for grad school, what you major in during college is almost irrelevant as long as you get good technical training (particularly in math & programming). From my perspective, undergrads are really not taught much but the most basic ideas.

BTW, i would not recommend some of the FAGA companies. I have friends in all of them and have talked to some of them myself. Apple notoriously ignore work-life balance. Amazon (i talked to them, mahy be, 2 years ago and I straight told them that i would not work for them, but may consider collaboration) put their scientists inside business units (so no labs) and they focuses a lot of business results instead of interesting science. FB is pretty good except they are mired in controversies and i do not know if that affect frontline engineers/scientists. Google is the place I would go.

In addition, working for bigger companies also means that it is harder to lead and do something meaningful. If I were you, i would consider start-ups too although sometimes it is also wise to work for a bigger company for a few years first.

Good luck and let me know if you have more questions.



回覆 引用 TOP

[隱藏]
引用:
原帖由 Iamsillyman 於 2019-6-14 11:11 AM 發表

Thank you for your sincere reply. I’m in the US either. I currently major in cog sci but I’m not really into it. I haven’t decided my dream job or career path but I wish I can work for FAGA in Silicon Valley. It is extremrly hard to declare my major as computer science, and one of my friend told me that the area of data science is closer to CS. However, I know nothing about it. Do you recommend I change my major to DS?
Good idea if you can switch to DS (or stat or CS)
another possible way is while keeping your cog sci major, and read a few important CS and Stat UG courses (possibly get a minor in CS/Stat/DS) in order to prepare yourself for grad school in DS-



回覆 引用 TOP

引用:
原帖由 Iamsillyman 於 2019-6-14 11:11 AM 發表


Thank you for your sincere reply. I’m in the US either. I currently major in cog sci but I’m not really into it. I haven’t decided my dream job or career path but I wish I can work  ...
我仔現在美国做data science工作, 我知有限, 只能簡單講吓.
佢, double major stat/CS, 美籍, 大學未畢業己經攞到world top firm offer, 佢話大把工做, 想先體驗生活, 2017年夏天入職起薪點是年薪USD82K + great benefit, 我見個job title係Technology Analyst. 做咗一年半, 今年初轉工去另一間top firm, 佢話better pay, 更重要既係新工作直接跟data science有關.  我聽佢意思, 再做一年半, 明年九月, 讀master, 目標data science.



回覆 引用 TOP

Statistics, Linear Algebra, multi-variable calculus  
/ Database
/ Big Data 
/ Machine Learning Algorithms 
/ Deep Learning 
/ Convex optimization. 
/ time series analysis , stochastic calculus



回覆 引用 TOP

Data science is basically data analysis, but you build model and use machine learning algorithm to help you solve problems, make predictions and visualize complicated data. It would include big data, data visualizations, machine learning, AI, computer vision, statistics, deep learning, reinforcement learning. I am working on a MSCS Machine learning track. It’s a lot of math.

Most of the real Data Scientists require PhD. Most people are just data analysts or data science practitioners.

[ 本帖最後由 thetaro 於 2019-6-19 02:07 AM 編輯 ]



回覆 引用 TOP

I am glad that so many are in the field.


One more point. You do not really need to jump right into a data scientist position if you want to pursue a career in the field. if you can do a PhD (both time & aptitude), it would be great. 


But there are other positions like data engineer which does not require a PhD. I have couple (as mentioned before) of good master students who get good data science related jobs (though not the top-tier data scientist positions in a silicon valley research lab).



回覆 引用 TOP

[隱藏]
提示: 作者被禁止或刪除 內容自動屏蔽
[按此隱藏 Google 建議的相符內容]