Clever Geek Handbook

Is Data Science a Bubble?



At SkillFactory, as a school that specializes in training data scientists and data analysts, we are attentive to the issue of the perception of the profession itself, both by the students themselves and by their employers. We have already talked about the requirements for the Data Analyst profession and the confusion in vacancies in this material , and now we want to share with you the translation of an article by the head of the intelligent decision making department at Google, in which she talks about the prospects for the Data Scientist position. About the company's risks in hiring a Data Scientist due to different understanding of the position or due to inexperienced HR and how to protect you from mistakes in your resume.





Is Data Science a Bubble? You'd be surprised how often I get asked this question. My answer?



Probably not, but the position of Data Scientist may be.



Let me explain myself before the pitchfork crowd arrives.



Data Scientist and the definition of data science



There are many opinions, but I prefer the following definition: " Data science is the discipline that makes data useful." If you don't like my definition, you might like Harlan Harris's clear definition:

Data science is defined by what a Data Scientist does and what he does is very well covered ... But who a Data Scientist is, perhaps a more fundamental question.



Okay. Who are we then? Well, it depends on which club you belong to. And this is where the bubble idea comes into play.



AND versus OR





For some people, the position implies full competence in three specializations (statistics, data collection, machine learning). For others, it means that the skills of one are combined with those of others. This different understanding of the position can hurt you when hiring.



Option one. Only worthy ones!



There are those who dream of stripping copycats of their jobs and limiting the sacred profession to an elite that understands everything about data.



How does it feel when such people are interviewing? They want to see me as a thoroughbred statistician, with a knowledge of machine learning, a black belt in analytics and a portfolio of applied projects. They want to know if my graduate school was trendy. They want to test how I was a leader and how I solved business problems. Oh, and I better have communication skills. What about the moon on a stick? If I hadn’t been doing data since I was eight, I would have been terrified. Until then, this is a fun little club that I have deeply conflicting feelings about.



Let me call it what it is: Club And... Participants must be competent statisticians, machine learning experts, and analysts with jewelery coding skills. Please note that getting into this club is quite difficult; not many people are experts on all data issues. These people will never be able to meet the world's needs for specialists. Sadly, this is the nature of supply and demand.



Option two. The doors are open to everyone!



An alternative and much more crowded club is the OR club . It consists of people who have renamed their job title, such as analyst or statistician, to a generic term. Sounds better, increases employment in data science, expands the community, brings a variety of skills. Everyone wins. Right? Almost.



What I like: It emphasizes the team, athletic nature of data science and this approach allows more people to participate in data science. Excellent! And some areas of data science aren't all that hard. Data mining, that is, collecting data, is something people are more skilled in than they think. If you thought collecting data required a PhD, I have good news. All you need is an understanding of how to look into datasets, moderate humility, and common sense.



What about the other side? Data science is renowned for its high demands and long immersion in learning. I deeply sympathize with the poor, bewildered hiring managers who think they are luring a versatile specialist but are hiring someone far less qualified. False advertising is damaging.



Tip: If you want complete assurance that you are not exaggerating on your resume, a Data Analyst position is the safest option.



False advertising



To be honest, every time I became a Data Scientist, I did the same job as before, in positions that were named differently. When the “effective managers” changed their job titles again, my responsibilities did not change in the slightest.



I'm not an exception. I have many ex- statisticians, decision support engineers, quantitative analysts, professors of mathematics, big data scientists, business analysts, leading analysts, research scientists, software engineers, PhDs ... all the proud modern scientists.



When I became Data Scientist, my responsibilities did not change at all.



Friends, I do not judge. Manage your personal brand well. But I want to point out that the definition of data science based on the concept of "Data Scientist" is not very accurate, given how motley the crowd attracts the name. Taking the limit, we get a set of words carefully crafted to say as little as possible. This is reflected in the way Data Scientist is seen. I recently felt a spike in pressure when a data science hiring manager posted something like “Do you have a PhD? Then you are probably a Data Scientist . " Rephrased to protect the innocent.



Using job titles to define data science is a dangerous game.



Well-known data scientists already know what they are looking for and can find a good, ahem, Data Scientist, even if the job title says a space alien . I'm worried about less experienced hiring managers. Many companies starting out in data science do not have the expertise to help them. Their plan? Hire a Data Scientist and everything will be fine.



Caution to those who pay



Put yourself in the shoes of a new hiring manager: you've read a lot and decided that you need skills in statistics, data collection, and machine learning for your project. You can hire three people. Now let's look at the candidates: 10 resumes labeled "Data Scientist".



If these are people of the I club , you can choose any three resumes. Each candidate has the skills you need. Unfortunately, this club is small (read: very expensive to hire ), so chances are good that these 10 people are not members of the I. It



can be difficult for the hiring manager to determine which part of data science a candidate is really good at.



If they are people OR(which is more likely in today's environment), you must interview them carefully to find out what they are really skilled at. You need three different skill sets. The people in front of you may only have one, but they also have every incentive to convince you that they are universal. They may have a minimum of knowledge about all three areas (statistics, data collection, machine learning). This can be dangerous for both your project and the hiring. You need to figure out what they are really good at, which can be difficult unless you are a seasoned data scientist with a variety of backgrounds.



Result? Mistakes when applying for a job. Resume buzzwords are not a guarantee of real skills.


I've seen many teams accidentally get multiple analysts to collect data instead of a diverse group. But this is not just a data science problem. It turns out that buzzwords on a resume don't necessarily come with skill guarantees. The hotter the buzzword, the more it spreads.



Is this the end of data analytics?



I personally treat job titles with a grain of salt. It is important that skills match all the needs of the company. If job title isn't a good indicator, then qualified hiring managers will learn to look for something else on their resume.



Lost in thought, I can even foresee the history of the OR club . The position may just go out of style, but I'm not the type to follow.



So is Data Science a bubble or not?



More data in the world means more demand for the three main data science activities - statistical inference, machine learning, analytics / data collection - so these skills will remain very relevant despite their names, which may change. You can always make a living by extracting value from data.



On the other hand, teams that have been hired on the hype and have never learned to focus on what's valuable to the business may find that their time is running out.



A few years ago, a friend of mine, a CTO who works in technology, complained about his useless data scientists. “I think you could hire data scientists like a drug lord buys a tiger for his estate,” I told him. "You don't know what you want from a tiger, but all the other drug lords have one."



I don't know real drug lords (or tigers), so I don't know what they have on their estates. But you understand me.



While Data Science sounds like a bubble, I'm actually optimistic. The growing volume of data means growing opportunities. All this requires good management. A friend of mine, for example, ended up solving many of his problems by realizing, with the help of data analysts, that part of his organization needed training. Since then, his teams have become more thoughtful about how to distribute work. Great things have begun. Teaching decision makers how to use data science saved the day!



Make sure the decision-makers have the right skills for the job. If a bubble exists, this could be its root.


The challenge for today's data science leaders is to help decision-makers get through this kind of training. This will create more people who can point out Data Scientist's technical talent in valuable directions. Read more here . Once data scientists bring value, their content becomes a necessity, not a fashion issue. Can we deal with the issues before the Data Scientist job loses popularity and the rebranding begins?





In our courses, we constantly analyze the market and give students those real skills that will allow them to remain professionally in demand for a long time. And in order to present the acquired skills correctly, we work with the graduate's resume and portfolio.



image





E






More articles:


All Articles