In this article, I present the courses and books that seem to me the most optimal for learning machine learning / data science from scratch. I try to give a list that will be as short as possible and at the same time give all the knowledge necessary to get started in practice, without serious knowledge gaps.
Disclaimer
You can understand what these recommendations are based on by reading the previous articles, which describe my path and principles of self-study, as well as general considerations on how to build the learning stages:
Previous articles
The guidelines in this article will become outdated, and for sure, there are now excellent courses and books that could be included in it. But these are at least some of the best materials on their topics. To prepare this list, dozens of courses and books were discarded, which are also aimed at learning from scratch, but are worse at presenting fundamental concepts.
The guidelines do not cover all potentially required technical skills. To get an idea of everything that is likely to need to be mastered - see Learning data science from scratch: milestones and milestones.
I am not citing materials about neural networks because, in most cases, I consider it ineffective to start learning with them, or study them at the early stages of self-learning.
Basic skills required
Knowledge of programming basics: Python and SQL
It is impossible to do machine learning or data science without knowing programming in Python or R (Better start with Python). Also, the vast majority of vacancies in "classic" machine learning (solving business problems, and working with initially numerical / statistical data) will require knowledge of SQL. For basic guidelines on how to learn them, see Self Learning in Data Science, From Zero to Senior in Two Years .
Maths
. , , , .
, - : ( ), (, , , ), , .
: . , 2 3 . 2 3 , . Math for machine learning, London Imperial College.
( , ), , : Robert Ghrist. . coursera.org, stepik.org
,
Datasmart ( ) - . , data science. , , . , python, pandas, scikit-learn - ( , .. ).
. . ( , .. , , ). - (). , .
— 8 , 2 — seaborn ( , ).
Data Science for Business: What You Need to Know about Data Mining and Data-Analytic Thinking, by Foster Provost ( ) - , , . Datasmart, . " " () . , ..
, - , / - . .
, , , . . , .. , , . . , .
" " - , . -, . , . Python, Pandas, scikit-learn.
- , Python. Applied DS with Python ( 1 3. 2, 4 )
, ( jupyter, pandas). , , . , - .
https://stepik.org/course/4852/syllabus
« » ( " " ). 3 , , 1 , . , ; . 4 , , , , .. , .
Kaggle : ( ) , , , ( .. ). - , , , .
scikit-learn - . , .. .
pandas, - , . , . , .
Python python standard library - . , . , collections itertools
-
( ). . , , : https://www.coursera.org/learn/competitive-data-science. kaggle - .
,
- . - -.
, - , Goodhart's law. .
/
, , https://towardsdatascience.com; 3 .
Statistics Done Wrong .The woefully complete guide by Alex Reinhart - . .
Python Machine Learning, by Sebastian Raschka - , . .
, - .
Git - . - Ry's Git tutorial. git. : http://ndpsoftware.com/git-cheatsheet.html
https://www.practicaldatascience.org/ - . , , , ( Cloud , .. , ).
?
. , , - , - /, , , , - , . - .
/
, , , .
, - self.development.mentor gmail.com,
Some realized that it was better for them to go to another area (programming, big date), some were able to adjust the curriculum / career plan for individual needs, to some I advised those who could help them better, and saved someone (?) From ineffective spending time for dead-end projects.
And if my articles are useful for you, you can also motivate me financially for future articles, under this article there should be a "donate" button for these purposes.