Notes about Machine Learning, Data Science and Analytics Engineering

How To Learn a Data Science – my experience

Overview of online courses to enter the Data Scientist profession. All courses are completed and a subjective view and experience is described. Only English language courses are included in this note.

How To Learn a Data Science
How To Learn a Data Science

Classic ML

HSE Advanced Machine Learning Specialization

The specialization consists of 7 courses. I only completed 2 and they are two different experiences. Course – How to Win a Data Science Competition: Learn from the Best Kagglers, I loved it. Don’t be confused by the name, the course gives you very practical things. For example, working with features, mean encoding, EDA and searching for leaks in data. Greatly expands your skillset and perspective on working with data. I highly recommend it. Course – Natural Language Processing, I got a neutral impression of it. The course gives NLP very extensively and deeply, but when doing assignments and tests, you have to google and read a lot. The course seemed difficult, and the goal was just to get into NLP, but do not forget that the specialization is called Advanced Machine Learning. I recommend it if you already have basic knowledge of NLP.

DataCamp 

The only paid platform in the review. Frequent discounts allow you to buy for $ 99 annual access to all courses. The platform has courses in Python, R, SQL. The site has career tracks, for example Data Scientist with Python. It was consists of 26 courses – Python, visualization, ml stack of Python libraries, classic ML, a little Deep Learning. The main advantage of courses on DataCamp is that you can quickly get skills of writing simple and basic Python code.

Deep Learning

Back-propagation
Back-propagation

EE-559 – Deep Learning course from École Polytechnique Fédérale de Lausanne, Switzerland

The course I started with Deep Learning and Pytorch. It covers all areas of work with neural networks: computer vision, NLP, GAN. The course consists of video materials, presentations, lectures and assignments on Pytorch. The material very good introduces the theory of machine learning and neural networks. In my opinion, this part is the most interesting and thoughtful part. There, a lot of attention is paid to mathematics and formulas. I highly recommend it.

AI for Medical Diagnosis Andrew Ng

What you shouldn’t expect from the course: an explanation of the basics of neural networks, a story about how to train best networks. Only in one task you will have write code to train the unet network, but it does not have to be run to complete the task. All exercises are writing code (Keras) into a ready-made jupiter notebooks in a marked place. Many things, such as GradCam, do not explain at all, but they just let you run it and see the result. On the plus side, there is a good section on model evaluation. Good, because they give you code that you can reuse yourself. I was too lazy to write this. If you have already taken courses or know how to use neural networks, then AI for Medical Diagnosis will be a good addition. Provided that the topic of medicine is interesting to you. Wouldn’t recommend.

Post in Russian with a list of courses

Share it

If you liked the article - subscribe to my channel in the telegram https://t.me/renat_alimbekov or you can support me Become a Patron!


Other entries in this category: