Statistics + Computer Science = Data Science? [closed]

Question

i want to become a data scientist. I studied applied statistics (actuarial science), so i have a great statistical background (regression, stochastic process, time series, just for mention a few). But now, I am going to do a master degree in Computer Science focus in Intelligent Systems.

Here is my study plan:

• Machine learning
• Advanced machine learning
• Data mining
• Fuzzy logic
• Recommendation Systems
• Distributed Data Systems
• Cloud Computing
• Knowledge discovery
• Business Intelligence
• Information retrieval
• Text mining

At the end, with all my statistical and computer science knowledge, can i call myself a data scientist? , or am i wrong?

Thanks for the answers.

Answer 1

I think that you're on the right track toward becoming an expert data scientist. Recently I have answered related question here on Data Science StackExchange (pay attention to the definition I mention there, as it essentially answers your question by itself, as well as to aspects of practicing software engineering and applying knowledge to solving real-world problems). I hope that you will find all that useful. Good luck in your career!

Answer 2

Well it depends on what kind of "Data Science" you wish to get in to. For basic analytics and reporting statistics will certainly help, but for Machine Learning and Artificial Intelligence then you'll want a few more skills

• Probability theory - you must have a solid background in pure probability so that you can decompose any problem, whether seen before or not, into probabilistic principles. Statistics helps a lot for already solved problems, but new and unsolved problems require a deep understanding of probability so that you can design appropriate techniques.

• Information Theory - this (relative to statistics) is quite a new field (though still decades old), the most important work was by Shannon, but even more important and often neglected note in literature is work by Hobson that proved that Kullback-Leibler Divergence is the only mathematical definition that truly captures the notion of a "measure of information". Now fundamental to artificial intellgence is being able to quantify information. Suggest reading "Concepts in Statistical Mechanics" - Arthur Hobson (very expensive book, only available in academic libraries).

• Complexity Theory - A big problem many Data Scientists face that do not have a solid complexity theory background is that their algorithms do not scale, or just take an extremely long time to run on large data. Take PCA for example, many peoples favourite answer to the interview question "how do you reduce the number of features in our dataset", but even if you tell the candidate "the data set is really really really large" they still propose various forms of PCA that are O(n^3). If you want to stand out, you want to be able to solve each problem on it's own, NOT throw some text book solution at it designed a long time ago before Big Data was such a hip thing. For that you need to understand how long things take to run, not only theoretically, but practically - so how to use a cluster of computers to distribute an algorithm, or which data structures take up less memory.

• Communication Skills - A huge part of Data Science is understanding business. Whether it's inventing a product driven by data science, or giving business insight driven by data science, being able to communicate well with both the Project and Product Managers, the tech teams, and your fellow data scientists is very important. You can have an amazing idea, say an awesome AI solution, but if you cannot effectively (a) communicate WHY that will make the business money, (b) convince your collegues it will work and (c) explain to tech people how you need their help to build it, then it wont get done.

Answer 3

Data scientist (to me) a big umbrella term. I would see a data scientist as a person who can proficiently use techniques from the fields of data mining, machine learning, pattern classification, and statistics.

However, those terms are intertwined to: machine learning is tied together with pattern classification, and also data mining overlaps when it comes finding patterns in data. And all techniques have their underlying statistical principles. I always picture this as a Venn diagram with a huge intersection.

Computer sciences is related to all those fields too. I would say that you need "data science" techniques to do computer-scientific research, but computer science knowledge is not necessarily implied in "data science". However, programming skills - I see programming and computer science as different professions, where programming is more the tool in order solve problems - are also important to work with the data and to conduct data analysis.

You have a really nice study plan, and it all makes sense. But I am not sure if you "want" to call yourself just "data scientist", I have the impression that "data scientist" is such a ambiguous term that can mean everything or nothing. What I want to convey is that you will end up being something more - more "specialized" - than "just" a data scientist.