What is the path to big data?

Question

Python

I am learning to program. The direction of big data is interesting. Promising direction. There are a couple of rather amateurish questions:
1. It is right that big data and machine learning specialists do not become from scratch. Only a programmer with at least a few years of experience can come to this technology?
2. Monitored vacancies by bigdata. There are slightly more vacancies for python programmers than for java. Is there a tendency to replace Java with Python in the big data niche? What language should be learned with an eye on big data in the future?

Answer 1

Read , remember and repeat before going to bed:

There is no universal "silver bullet" against the challenges of the most complex projects, including big data processing.
My project architecture, including the architecture of the database used for this project, is broadly and completely dependent on functional requirements!
I understand this, every time I will independently design the architecture of the project, including the data storage system (SHD) for this project.
And finally: no matter how much data and what type, it is important - how they are connected!

Answer 2

I like the conditional plan written here .
Regarding "preemption" as you put it in Java Python in this context, it's fair.
No advantages of Java for data analysis, whether large or small, are needed in most cases.

Answer 3

there is no preemption, python is not able to supplant java, from data analysis, things are quite the opposite, for high-speed work, and for manipulating data in memory, it is statically typed languages ​​that are very popular java / scala.
all sorts of pythons and js are used only for prototyping, quick testing of ideas.
In addition, python is used abroad to teach students programming (HE "programming" specialties) as an easier language to learn, plus it is convenient to work using RELP ( https://ru.wikipedia.org/wiki/REPL ).
It is much easier to write something simple in dynamically typed languages. And that's why there are so many student crafts in python for data processing.
But keep in mind that the use of dynamically typed languages ​​for really large projects is deadly, both in terms of development and debugging, and in terms of performance.
For prototyping, for math controls, python is a great solution, for a real business project you shouldn’t even bother.
By the way, for those who like to assemble something quickly on their knees in java 9, jshell with relp will go https://blogs.oracle.com/java/entry/jshell_and_rel...

Answer 4

To the good answers above, I will add a link to such a very popular scheme called "Road To Data Scientist". You can print it out, hang it up, and tick off what you know you don't. It is clear that the scheme is quite relative, but still.

Answer 5

https://habrahabr.ru/company/npl/blog/321918/