Best Ways To Learn Data Science
By George Bennett
In this post I will share ways to learn and/or brush up on data science skills. I want to first say that, in my opinion, the best ways to learn data science are to attend college or coding boot camps. That being said, this field is always evolving and in order to stay competitive it is best to always be on the lookout to keep your skills sharp.
There are several resources available. Some are better for beginners and others are more advanced. Also preference plays a big role here. Different people learn differently, and what is the most effective choice for one may not be ideal for another. I learned my data science skills in a boot camp, but these resources absolutely enhanced my knowledge. The types of resources are as follows: coding practice sites, YouTube videos, books, Kaggle kernels, and blogs.
Coding practice sites such as codecademy.com, hackerrank.com, and others are a good place to start if you are brand new. You can learn to code here as well as practice. The main coding languages of data science are python and R. You will need at least one of these to do projects. In addition to one of those languages you will also need to learn SQL. It is essential for obtaining data out of relational databases. In particular hackerrank.com is a great site to practice your skills and keep them fresh.
YouTube videos are also good for beginners. Their are channels such as data school that will teach you as well as give you a road map of what you need to learn. YouTube videos are a good place to start, but one down side is that they can be very generalized and not go into details. Also if all you need is one bit of information, such as a mathematical formula or a method call, it can be hard to find.
Books are like bigger and badder YouTube videos. They give a road map and often come close to containing everything a beginner needs. Books can go into much more detail than YouTube videos, and if you need to reference a small bit of information, you can easily flip through a chapter and find it. One down side of books is that they cost money, but you can buy them used, or perhaps even find them in a library. One book that helped me was Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow by Aurelien Geron.
Once you learn the basics it is time to start doing projects. There is a plethora of data sets on kaggle.com. If you happen to get stuck while doing your project, or perhaps you think there is a better way to do it, simply check the kaggle kernels associated with the data set. They will be under the notebooks tab. Often these kernels are helpful, especially when doing a new style of project. The other resources may be ranked by preference, but I believe Kaggle projects are irreplaceable. Always remember to comment your code and avoid copying someone else’s code. One down side of the Kaggle kernels are that they are sometimes not commented correctly, and sometimes there are mistakes. This can make them hard to follow or lead to improper practices. Not all kernels are like this though, so check the 2nd or 3rd best kernels along the way.
The final resource is blogs. Blogs excel at answering questions quickly. Often when I get confused or forget something I will google it. This usually leads to a blog post, either on towards data science or on stack exchange. This method answers questions quickly and is a good way to patch up small holes in your knowledge.
These are the top resources for learning data science. If you are brand new, I would recommend learning python on one of the coding websites as a start, and then look into a YouTube playlist to learn how to work with pandas, numpy, and matplotlib. If you are just looking to brush up, then I would check the top blogs, kaggle kernels, and practice coding.
I would like to reiterate that the best ways to learn data science are either through college or a good boot camp. Although it may be beneficial to get your feet wet in the material to see how you like it before you invest.