June 25, 2019 at 10:23“And if I don't know math, am I hopeless?” - specialists answer frequently asked questions about professions in Data Science
Interest in Data Science continues to grow, the market needs good specialists. But the threshold for entering the profession is quite high, newcomers are often stopped by myths and stereotypes about the field - “it’s long, difficult, it is better not to meddle without physical education”. We collected the most common questions and concerns of those who start a career in Data Science and asked specialists to answer them.
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis" The
question about mathematics is ambiguous. A thorough knowledge of mathematics is neither a necessary nor sufficient condition. Of course, those who know her will be easier. But all the necessary knowledge is given either in the classroom or in additional materials.
Here, as in sports. There are people who can run a marathon without preparation. The rest will be harder, but with enough preparation, they will run. The math base is cool, but not critical.
Daria Mukhina, Skyeng product analyst, Netology analytics course consultant
It seems that now a deep mathematical base can be replaced by the ability to google. There are a lot of videos and articles on the Internet where you can get the information presented in an accessible way - and you don’t need to go into university textbooks. The main thing is to know what you need.
Now more important is the ability to apply knowledge in a real task, and not just possess it.
Elena Gerasimova, Head of Data Science in Netology,
The concept of "specialized technical or mathematical education" is a thing of the past. Those who are confident in their skills and domain knowledge will not be compared with a graduate from the Moscow Institute of Physics and Technology with a knowledge of mathematics — they will be compared by the usefulness of a business for solving problems.
Dozens of working algorithms and libraries are already known that are capable of taking on the entire mathematical part without human intervention.
Konstantin Bashevoy, an analyst and developer at Yandex and a teacher of the Python for Data Analysis
course . Of course, the easiest way to enter the DS sphere is for those people who have experience in training or working in a technical field.
Although the division into “techies” and “humanities” is very arbitrary, Data Scientist needs mathematics, not grade 8, but higher. You can study everything yourself, but if a person graduated from a technical university - most likely, he already has the necessary base. Those who have programming experience and understanding of algorithms will also find it easier. If Python is very difficult for a person, it will be more difficult for him - after all, they will begin to talk about probability theory, then about neural networks.
The experience of studying at a physical laboratory or working in engineering specialties greatly simplifies the development of DS. However, one must remember that there are still a huge number of near-DS specializations that you can come to without a deep knowledge of mathematics. It’s not necessary to be a Data Scientist, with a good understanding of the business, you can become an excellent BI analyst.
Alexey Kuzmin, development manager at DomKlik, Data Scientist, teacher of machine learning courses in Netology.
It all depends on the task. This is really a difficult choice, there is no ready-made recipe. I would take a developer - for my company’s tasks, such a profile is closer.
Konstantin Bashevoy, an analyst-developer at Yandex and a teacher of the course "Python for data analysis"
And we have more mathematics analytics. But in general, everything really depends on the task. If the employer has a highly loaded banking service, then he most likely needs a developer who quickly closes a large number of technical tasks and helps with DS and models. If the company has a project that is already set up and working smoothly, then junior employees may be suitable for its support.
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis"
Of course! High places at Kaggle is a great portfolio project. Sometimes the platform is criticized for "idealized" conditions. Of course, there is no platform fault in this. Usually, when a task is set for a scientist or analyst, it does not begin with building cool models, but with managerial work, preparing data and tools. Where to get the necessary data? How to handle all this? What are the unobvious problems in the data? This part on Kaggle is usually not.
When they made the model, another stage begins - implementation. In addition to the fact that the system should work in prod, you need to prove its value for the business, teach your colleagues how to use it and, possibly, “sell” it to the customer.
Therefore, sometimes an employee builds cool models, but in real conditions has difficulty with the first and third part of the work. If a person has good communication skills, he has excellent programming abilities and, in addition, builds accurate models - he has no price. At Kaggle, you hone model building, but you will need a lot of applied skills to apply this in real projects.
Alexey Kuzmin, development manager at DomKlik, Data Scientist, teacher of machine learning courses in Netology.
Everything depends very much on the tasks and on the profile of the company. If this is a startup for 5 people, then an analyst who knows how to deal with personnel can come in handy simply because the startup has no people for the personnel. If this is a large, serious, large company with projects that last for years, in which the same people do the same tasks, then you will need a narrow specialist who knows only one specific area and nothing more.
Soft skills for communication skills, stress tolerance, working capacity, and ability to understand the applied field are a separate plus.
It is useful when a specialist has the skills to work with business, then it is easier for him to understand the requests and tasks of the company, he can plunge into the problem and offer some alternative solution.
In addition, there is now a huge shortage of specialists in the market with skills related to DS. For example, for a very long time we searched for a Product Owner with an understanding of DS, so that he could create products that were based on artificial intelligence.
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis" The
advice is commonplace - go to interviews. Often written in vacancies is different from the real requests of the employer. The interview gives you the opportunity to find out on which project the work is planned, what tools will need to be used and with what people to work. My advice is to take the text of the vacancy as a guideline, and not the ultimate truth.
Adequate employers understand that if you worked with Google Cloud, and they use Azure, then this is not a problem - the specialist will quickly relearn. There are much more important things: what exactly you will have to do, how the processes in the team are arranged - this can be found out only in person. In vacancies, such details do not indicate.
Elena Gerasimova, Head of Data Science in Netology,
Remote work in similar positions in large IT companies is really the exception rather. Nevertheless, many foreign companies with Russian representative offices are ready for the sake of saving on salaries and relocating to a remote format when performing assigned tasks.
Start-ups are also often looking for remote employees - if it is the udalenka that matters, it is worth looking for such vacancies.
In general, I think that working in the office for analysts and Data Scientist is preferable - without working in the office, you are depriving yourself of the opportunity to study with colleagues right at the workplace, communicate with the team, quickly resolve issues (well, and take advantage of a good office: a gym, dinners, a change of scenery).
Konstantin Bashevoi, an analyst and developer at Yandex and a teacher of the Python for Data Analysis course
. We had guys who, after 30, moved from industrial professions to developers: it turned out that everything in the department was 5-8 years younger - but these were trifles.
Of course, if a person moves to DS at the age of 65, then yes, probably, it will be hard for him. And so there is a huge number of cases when people moved to DS from very remote areas, for example, medicine, aged 30-40 years.
Another important point - when moving into a new sphere, you must be prepared to lower salaries. If the specialist has a family and three children, it will be stressful. In general, there are a lot of positive examples, and the level of salary is growing in parallel with new experience.
Elena Gerasimova, Head of Data Science in Netology
When moving to DS in adulthood, the spirit and willingness to sacrifice some of your established principles and accept the rules of the game that are provided for in this environment are extremely important. We recently graduated with honors from a student with three children: he took care leave during his studies, and his wife worked during this period. He really wants to be a Scientist date, a very talented graduate and his motivation is stronger than the surrounding circumstances.
Daria Mukhina, Skyeng product analyst, Netology analytics courses consultant
For any person, the question of salary for an interview is a stressful question. I think trying to somehow joke on this topic or evade is risky. It is best to conduct a mini-study before the conversation, to unload the vacancies where the plug is indicated: upper threshold, lower threshold. Understand how much you need, conditionally, money for living, then once again look at salary forks at the level of June - and name the amount that will fit into them, but will not be lower than your subsistence level.
“What math is needed? If there is no matbase, am I hopeless? ”
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis" Thequestion about mathematics is ambiguous. A thorough knowledge of mathematics is neither a necessary nor sufficient condition. Of course, those who know her will be easier. But all the necessary knowledge is given either in the classroom or in additional materials.
Here, as in sports. There are people who can run a marathon without preparation. The rest will be harder, but with enough preparation, they will run. The math base is cool, but not critical.
Daria Mukhina, Skyeng product analyst, Netology analytics course consultant It seems that now a deep mathematical base can be replaced by the ability to google. There are a lot of videos and articles on the Internet where you can get the information presented in an accessible way - and you don’t need to go into university textbooks. The main thing is to know what you need.
Now more important is the ability to apply knowledge in a real task, and not just possess it.
Elena Gerasimova, Head of Data Science in Netology,The concept of "specialized technical or mathematical education" is a thing of the past. Those who are confident in their skills and domain knowledge will not be compared with a graduate from the Moscow Institute of Physics and Technology with a knowledge of mathematics — they will be compared by the usefulness of a business for solving problems.
Dozens of working algorithms and libraries are already known that are capable of taking on the entire mathematical part without human intervention.
“Well, and what kind of background is easier to enter the DS sphere? Obviously, this is math, but what else will help? ”
Konstantin Bashevoy, an analyst and developer at Yandex and a teacher of the Python for Data Analysiscourse . Of course, the easiest way to enter the DS sphere is for those people who have experience in training or working in a technical field.
Although the division into “techies” and “humanities” is very arbitrary, Data Scientist needs mathematics, not grade 8, but higher. You can study everything yourself, but if a person graduated from a technical university - most likely, he already has the necessary base. Those who have programming experience and understanding of algorithms will also find it easier. If Python is very difficult for a person, it will be more difficult for him - after all, they will begin to talk about probability theory, then about neural networks.
The experience of studying at a physical laboratory or working in engineering specialties greatly simplifies the development of DS. However, one must remember that there are still a huge number of near-DS specializations that you can come to without a deep knowledge of mathematics. It’s not necessary to be a Data Scientist, with a good understanding of the business, you can become an excellent BI analyst.
“And who is still preferable for the employer: a person with knowledge of Python and a background for a developer, or a graduate with strong mathematics?”
Alexey Kuzmin, development manager at DomKlik, Data Scientist, teacher of machine learning courses in Netology.It all depends on the task. This is really a difficult choice, there is no ready-made recipe. I would take a developer - for my company’s tasks, such a profile is closer.
Konstantin Bashevoy, an analyst-developer at Yandex and a teacher of the course "Python for data analysis"And we have more mathematics analytics. But in general, everything really depends on the task. If the employer has a highly loaded banking service, then he most likely needs a developer who quickly closes a large number of technical tasks and helps with DS and models. If the company has a project that is already set up and working smoothly, then junior employees may be suitable for its support.
“Should Kaggle be seen as an aid to entering DS? Are employers looking at Kaggle Masters? ”
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis"Of course! High places at Kaggle is a great portfolio project. Sometimes the platform is criticized for "idealized" conditions. Of course, there is no platform fault in this. Usually, when a task is set for a scientist or analyst, it does not begin with building cool models, but with managerial work, preparing data and tools. Where to get the necessary data? How to handle all this? What are the unobvious problems in the data? This part on Kaggle is usually not.
When they made the model, another stage begins - implementation. In addition to the fact that the system should work in prod, you need to prove its value for the business, teach your colleagues how to use it and, possibly, “sell” it to the customer.
Therefore, sometimes an employee builds cool models, but in real conditions has difficulty with the first and third part of the work. If a person has good communication skills, he has excellent programming abilities and, in addition, builds accurate models - he has no price. At Kaggle, you hone model building, but you will need a lot of applied skills to apply this in real projects.
“What competencies, in addition to technical competencies, are needed for a novice specialist so that the employer will notice him among the general stream?”
Alexey Kuzmin, development manager at DomKlik, Data Scientist, teacher of machine learning courses in Netology.Everything depends very much on the tasks and on the profile of the company. If this is a startup for 5 people, then an analyst who knows how to deal with personnel can come in handy simply because the startup has no people for the personnel. If this is a large, serious, large company with projects that last for years, in which the same people do the same tasks, then you will need a narrow specialist who knows only one specific area and nothing more.
Soft skills for communication skills, stress tolerance, working capacity, and ability to understand the applied field are a separate plus.
It is useful when a specialist has the skills to work with business, then it is easier for him to understand the requests and tasks of the company, he can plunge into the problem and offer some alternative solution.
In addition, there is now a huge shortage of specialists in the market with skills related to DS. For example, for a very long time we searched for a Product Owner with an understanding of DS, so that he could create products that were based on artificial intelligence.
“How to navigate vacancies and are not afraid if new tools are indicated for you?” What does it take to go and start working in a profession? ”
Konstantin Bashevoi, an analyst-developer at Yandex and a teacher of the course "Python for data analysis" Theadvice is commonplace - go to interviews. Often written in vacancies is different from the real requests of the employer. The interview gives you the opportunity to find out on which project the work is planned, what tools will need to be used and with what people to work. My advice is to take the text of the vacancy as a guideline, and not the ultimate truth.
Adequate employers understand that if you worked with Google Cloud, and they use Azure, then this is not a problem - the specialist will quickly relearn. There are much more important things: what exactly you will have to do, how the processes in the team are arranged - this can be found out only in person. In vacancies, such details do not indicate.
“Is it true that there is no remote work in the DS market?”
Elena Gerasimova, Head of Data Science in Netology,Remote work in similar positions in large IT companies is really the exception rather. Nevertheless, many foreign companies with Russian representative offices are ready for the sake of saving on salaries and relocating to a remote format when performing assigned tasks.
Start-ups are also often looking for remote employees - if it is the udalenka that matters, it is worth looking for such vacancies.
In general, I think that working in the office for analysts and Data Scientist is preferable - without working in the office, you are depriving yourself of the opportunity to study with colleagues right at the workplace, communicate with the team, quickly resolve issues (well, and take advantage of a good office: a gym, dinners, a change of scenery).
“And what if at the age of 40 I become a junior Scientist? What are my prospects? Where am I and how to move? ”
Konstantin Bashevoi, an analyst and developer at Yandex and a teacher of the Python for Data Analysis course. We had guys who, after 30, moved from industrial professions to developers: it turned out that everything in the department was 5-8 years younger - but these were trifles.
Of course, if a person moves to DS at the age of 65, then yes, probably, it will be hard for him. And so there is a huge number of cases when people moved to DS from very remote areas, for example, medicine, aged 30-40 years.
Another important point - when moving into a new sphere, you must be prepared to lower salaries. If the specialist has a family and three children, it will be stressful. In general, there are a lot of positive examples, and the level of salary is growing in parallel with new experience.
Elena Gerasimova, Head of Data Science in NetologyWhen moving to DS in adulthood, the spirit and willingness to sacrifice some of your established principles and accept the rules of the game that are provided for in this environment are extremely important. We recently graduated with honors from a student with three children: he took care leave during his studies, and his wife worked during this period. He really wants to be a Scientist date, a very talented graduate and his motivation is stronger than the surrounding circumstances.
“How can a novice specialist answer questions about salaries at an interview? How to evaluate yourself? ”
Daria Mukhina, Skyeng product analyst, Netology analytics courses consultant For any person, the question of salary for an interview is a stressful question. I think trying to somehow joke on this topic or evade is risky. It is best to conduct a mini-study before the conversation, to unload the vacancies where the plug is indicated: upper threshold, lower threshold. Understand how much you need, conditionally, money for living, then once again look at salary forks at the level of June - and name the amount that will fit into them, but will not be lower than your subsistence level.
From the editors
- "Profession the Data Scientist »
- Online course « Big the Data »
- Online course " Machine Learning "
- Online course « the Python for working with data "